Subscribe
Share
Share
Embed
The Question is a collaborative learning podcast about Design Systems. Smart people like you sign up, answer a few niche questions about design systems for each episode, and then we all get together to unpack the data we've gathered. Each week, I'll invite a new co-host to help facilitate the conversation. After the deep dive, the co-host and I record a recap of what we learned. That means, for each episode, you can listen to the recap and the full deep dive!
If you're a design system practitioner, subscribe today (https://bencallahan.com/the-question) to receive an invitation to each episode. This only works if the community joins in!
Stay in learning mode ❤️
Ben Callahan (00:04)
Hello, system thinkers. Welcome to The Question is a live collaborative learning show committed to broadening your perspective on design systems. I'm your host, Ben Callahan, founder of Sparkbox and Redwood's Design System Community. If you have a moment, please get subscribed to the show out at bencallahan.com/thequestion Let's get into it.
ep073 screen raw (00:29)
Davy, give us a quick overview of just like who you are and sort of what got you into the design system space. yeah. So I've been a product designer. So I'm one of the product designers on the Atlassian Design System previously at Meta, as you see on my hat in that portrait. I've been in design systems. ⁓ shoot, a while. Some number under 10 years, but over six years.
Ben Callahan (00:29)
Davy, give us a quick overview of just like who you are and sort of what got you into the design system space. yeah. So I've been a product designer. So I'm one of the product designers on the Atlassian Design System previously at Meta, as you see on my hat in that portrait. I've been in design systems. ⁓ shoot, a while. Some number under 10 years, but over six years.
ep073 speaker raw (00:29)
Davy, give us a quick overview of just like who you are and sort of what got you into the design system space. yeah. So I've been a product designer. So I'm one of the product designers on the Atlassian Design System previously at Meta, as you see on my hat in that portrait. I've been in design systems. ⁓ shoot, a while. Some number under 10 years, but over six years.
ep073 screen raw (00:57)
specifically and I think the compartmentalization, ⁓ production, operational aspects of this is tuned very well and then I think as ⁓ in the last nine months or so as things have been shifting and workflows have been changing, being able to systemize that sort of thing and repeat things in a much more rapid manner that has been a big
Ben Callahan (00:57)
specifically and I think the compartmentalization, ⁓ production, ⁓ operational aspects of this is tuned very well and then I think as ⁓ in the last nine months or so as things have been shifting and workflows have been changing, being able to systemize that sort of thing and repeat things in a much more rapid manner that has been a big
ep073 speaker raw (00:57)
specifically and I think the compartmentalization, ⁓ production, ⁓ operational aspects of this is tuned very well and then I think as ⁓ in the last nine months or so as things have been shifting and workflows have been changing, being able to systemize that sort of thing and repeat things in a much more rapid manner that has been a big
ep073 screen raw (01:27)
Interest of mine. I have a podcast called design system office hours linked there and we're recently Going to episode 100 so we're getting that's gonna be one of the next one. So that's like a pretty big deal based out of Sacramento home of the deftones and the Sacramento Kings and so those are two things We are known for Dude the podcast man. Everybody knows your voice and now they get to see your smiling face
Ben Callahan (01:27)
Interest of mine. I have a podcast called design system office hours linked there and we're recently ⁓ Going to episode 100. So we're getting that's gonna be one of the next one. So that's like a pretty big deal based out of Sacramento home of the deftones and the Sacramento Kings and so those are two things ⁓ We are known for Dude the podcast man. Everybody knows your voice and now they get to see your smiling face
ep073 speaker raw (01:27)
Interest of mine. I have a podcast called design system office hours linked there and we're recently ⁓ Going to episode 100. So we're getting that's gonna be one of the next one. So that's like a pretty big deal based out of Sacramento home of the deftones and the Sacramento Kings and so those are two things ⁓ We are known for Dude the podcast man. Everybody knows your voice and now they get to see your smiling face
thank you. Thank you. Second time. Two timer. Two time question co-host. That's right. That's right. Yeah. Every time we get a chance to chat, Davy, I'm just always I just always enjoy it, man. You have such a such a wonderful posture in the space. You just like welcoming and you challenge folks in like a really, I don't know, just like a gentle way that I think is needed in the space. So appreciate you coming on again, man. Thank you.
ep073 screen raw (01:56)
thank you. Thank you. Second time. Two timer. Two time question co-host. That's right. That's right. Yeah. Every time we get a chance to chat, Davy, I'm just always I just always enjoy it, man. You have such a such a wonderful posture in the space. You just like welcoming and you challenge folks in like a really, I don't know, just like a gentle way that I think is needed in the space. So appreciate you coming on again, man. Thank you.
Ben Callahan (01:56)
thank you. Thank you. Second time. Two timer. Two time question co-host. That's right. That's right. Yeah. Every time we get a chance to chat, Davy, I'm just always I just always enjoy it, man. You have such a such a wonderful posture in the space. You just like welcoming and you challenge folks in like a really, I don't know, just like a gentle way that I think is needed in the space. So appreciate you coming on again, man. Thank you.
ep073 screen raw (02:22)
OK, so when we were chatting, Davy, we knew that you I knew you wanted to talk AI in some capacity. And so ⁓ we landed on this concept of AI as automation in design systems. And we thought we would dig in in a couple of different ways. And so we asked four questions of all of you. The first question was, when you think about your current design systems workflow, what percentage of your time do you think could be automated?
Ben Callahan (02:22)
OK, so when we were chatting, Davy, we knew that you I knew you wanted to talk AI in some capacity. And so ⁓ we landed on this concept of AI as automation in design systems. And we thought we would dig in in a couple of different ways. And so we asked four questions of all of you. The first question was, when you think about your current design systems workflow, what percentage of your time do you think could be automated?
ep073 speaker raw (02:22)
OK, so when we were chatting, Davy, we knew that you I knew you wanted to talk AI in some capacity. And so ⁓ we landed on this concept of AI as automation in design systems. And we thought we would dig in in a couple of different ways. And so we asked four questions of all of you. The first question was, when you think about your current design systems workflow, what percentage of your time do you think could be automated?
ep073 screen raw (02:51)
And we gave you 0, 25, 50, 70. Oh, that should be 75, I think, and 100%. And an NA, of course. And the next question was, should be automated? So could versus should was sort of the first two questions. And then we asked you two follow-ups. The third question was, in what areas of our work do you believe we should avoid AI automation and why?
Ben Callahan (02:51)
And we gave you 0, 25, 50, 70. Oh, that should be 75, I think, and 100%. And an NA, of course. And the next question was, should be automated? So could versus should was sort of the first two questions. And then we asked you two follow-ups. The third question was, in what areas of our work do you believe we should avoid AI automation and why?
ep073 speaker raw (02:51)
And we gave you 0, 25, 50, 70. Oh, that should be 75, I think, and 100%. And an NA, of course. And the next question was, should be automated? So could versus should was sort of the first two questions. And then we asked you two follow-ups. The third question was, in what areas of our work do you believe we should avoid AI automation and why?
ep073 screen raw (03:14)
And the fourth question was, in our world, craft used to mean beautifully and elegantly designed and implemented interface work. In a 2026 design systems context, what does craft mean to you? And so we got so many funny answers. You guys know this every time we do it, it's just, get, y'all take so much time and care and thought when you answer. We sent this out to 1,077 design system practitioners. got 101 responses, crossed over that hundred threshold. Davy, that's awesome.
Ben Callahan (03:14)
And the fourth question was, in our world, craft used to mean beautifully and elegantly designed and implemented interface work. In a 2026 design systems context, what does craft mean to you? And so we got so many funny answers. You guys know this every time we do it, it's just, get, y'all take so much time and care and thought when you answer. We sent this out to 1077 design system practitioners. got 101 responses crossed over that hundred threshold. Davy, that's awesome.
ep073 speaker raw (03:14)
And the fourth question was, in our world, craft used to mean beautifully and elegantly designed and implemented interface work. In a 2026 design systems context, what does craft mean to you? And so we got so many funny answers. You guys know this every time we do it, it's just, get, y'all take so much time and care and thought when you answer. We sent this out to 1077 design system practitioners. got 101 responses crossed over that hundred threshold. Davy, that's awesome.
ep073 screen raw (03:44)
And as always, you can see the raw data here, do your own analysis and share that back. I would love to hear what you see in the data. ⁓ I loved this. I wanted to start, if it's okay, Davy, I want to start with this quote by Greg. I just thought this was like a nice framing for the day. ⁓ I think this was in the, I think this was in the fourth question, but he said,
Ben Callahan (03:44)
⁓ And as always, you can see the raw data here, do your own analysis and share that back. I would love to hear what you see in the data. ⁓ I loved this. I wanted to start, if it's okay, Davy, I wanted to start with this quote by Greg. I just thought this was like a nice framing for the day. ⁓ I think this was in the, I think this was in the fourth question, but he said,
ep073 speaker raw (03:44)
⁓ And as always, you can see the raw data here, do your own analysis and share that back. I would love to hear what you see in the data. ⁓ I loved this. I wanted to start, if it's okay, Davy, I wanted to start with this quote by Greg. I just thought this was like a nice framing for the day. ⁓ I think this was in the, I think this was in the fourth question, but he said,
ep073 screen raw (04:09)
There's always been a split between the people who believe in their craft and those that do this as a job. Neither is a slight on either group, but AI is empowering the latter. But the people who care about craft are the ones driving conversations like these and keeping track of the guardrails. I just thought that was a nice sort of like way to frame our conversation. know, nobody here I think is a hundred percent anti AI or tooling, right? Like we want to see the tools evolve.
Ben Callahan (04:09)
There's always been a split between the people who believe in their craft and those that do this as a job. Neither is a slight on either group, but AI is empowering the latter. But the people who care about craft are the ones driving conversations like these and keeping track of the guardrails. I just thought that was a nice sort of like way to frame our conversation. know, nobody here I think is a hundred percent anti AI or tooling, right? Like we want to see the tools evolve.
ep073 speaker raw (04:09)
There's always been a split between the people who believe in their craft and those that do this as a job. Neither is a slight on either group, but AI is empowering the latter. But the people who care about craft are the ones driving conversations like these and keeping track of the guardrails. I just thought that was a nice sort of like way to frame our conversation. know, nobody here I think is a hundred percent anti AI or tooling, right? Like we want to see the tools evolve.
ep073 screen raw (04:37)
But we also just want to be smart about how we use those things and where we put them into our workflows. ⁓ Those first two questions. So, Davy, I want to hear your sort of gut response to looking at the two charts here that are the more quantitative questions and sort of graphed out answers. What did you see when you looked at this? Yeah, I think the interest in this topic
Ben Callahan (04:37)
But we also just want to be smart about how we use those things and where we put them into our workflows. ⁓ Those first two questions. So, Davy, I want to hear your sort of gut response to looking at the two charts here that are the more quantitative questions and sort of graphed out answers. What did you see when you looked at this? Yeah, I think the interest in this topic
ep073 speaker raw (04:37)
But we also just want to be smart about how we use those things and where we put them into our workflows. ⁓ Those first two questions. So, Davy, I want to hear your sort of gut response to looking at the two charts here that are the more quantitative questions and sort of graphed out answers. What did you see when you looked at this? Yeah, I think the interest in this topic
ep073 screen raw (04:59)
Overall, so the last episode that we posted of the podcast was a ⁓ episode about the zero height design system report, which someone should help me link here. I'm going to post, have the link, but I don't have the figure out. ⁓ But ⁓ one of the questions there was ⁓ how many of y'all use automation in like your design token pipeline and 60 % said, no, like no, which this seems like. ⁓
Ben Callahan (04:59)
Overall, so the last episode that we posted of the podcast was a ⁓ episode about the zero height design system report, which someone should help me link here. I'm going to post, have the link, but I don't have the figure. ⁓ But ⁓ one of the questions there was ⁓ how many of y'all use automation in like your design token pipeline and 60 % said, no, like no, which this seems like. ⁓
ep073 speaker raw (04:59)
Overall, so the last episode that we posted of the podcast was a ⁓ episode about the zero height design system report, which someone should help me link here. I'm going to post, have the link, but I don't have the figure out. ⁓ But ⁓ one of the questions there was ⁓ how many of y'all use automation in like your design token pipeline and 60 % said, no, like no, which this seems like. ⁓
Ben Callahan (05:28)
when I was talking to my co-host PJ about it, design systems is ripe for automation. Like that's our job is to create repeatable processes. that 60 % that haven't done token automation. Then we start to think like, why haven't you done that? Are you interested in like the craft of ⁓ just manually curating these things? ⁓
ep073 screen raw (05:28)
when I was talking to my co-host PJ about it, design systems is ripe for automation. Like that's our job is to create repeatable processes. that 60 % that haven't done token automation. Then we start to think like, why haven't you done that? Are you interested in like the craft of ⁓ just manually curating these things? ⁓
ep073 speaker raw (05:28)
when I was talking to my co-host PJ about it, design systems is ripe for automation. Like that's our job is to create repeatable processes. that 60 % that haven't done token automation. Then we start to think like, why haven't you done that? Are you interested in like the craft of ⁓ just manually curating these things? ⁓
ep073 screen raw (05:55)
Are you interested in manually updating them? How much of this is more or less like a resistance to let go of things that ⁓ might be better suited for an agent to do? ⁓ Token automation specifically, ⁓ diffing them, manually updating them, I think is ripe for error. No matter how good you are with
Ben Callahan (05:55)
Are you interested in manually updating them? How much of this is more or less like a resistance to let go of things that ⁓ might be better suited for an agent to do? ⁓ Token automation specifically, ⁓ diffing them, manually updating them, I think is ripe for error. No matter how good you are with
ep073 speaker raw (05:55)
Are you interested in manually updating them? How much of this is more or less like a resistance to let go of things that ⁓ might be better suited for an agent to do? ⁓ Token automation specifically, ⁓ diffing them, manually updating them, I think is ripe for error. No matter how good you are with
ep073 screen raw (06:23)
typing things or copying and pasting things, right? Depending on how much sleep you've gotten that day or if you're doing this like while you're on a call, you might have one or two errors and that's like commonplace. Yeah. Yeah. Yeah. And I guess it's interesting because I don't know, just to challenge what you're saying, Davy, like, are you saying that ⁓ there would never be errors if we ask AI to do something like that? No, there I think there's certain tasks that
Ben Callahan (06:23)
typing things or copying and pasting things, right? Depending on how much sleep you've gotten that day or if you're doing this while you're on a call, you might have one or two errors and that's commonplace. Yeah. Yeah. And I guess it's interesting because I don't know, just to challenge what you're saying, Davy, are you saying that ⁓ there would never be errors if we ask AI to do something like that? No, I think there's certain tasks that
ep073 speaker raw (06:23)
typing things or copying and pasting things, right? Depending on how much sleep you've gotten that day or if you're doing this while you're on a call, you might have one or two errors and that's commonplace. Yeah. Yeah. And I guess it's interesting because I don't know, just to challenge what you're saying, Davy, are you saying that ⁓ there would never be errors if we ask AI to do something like that? No, I think there's certain tasks that
ep073 screen raw (06:53)
we do as design system maintainers that are really well suited for automation or placed somewhere in the pipeline for us to do and then come back and verify and then token automation and token linting and checking and diffing I think is one of those. Yeah, that makes sense. When I looked at these two answers, it's funny because if you just look at the charts, they're almost identical, right?
Ben Callahan (06:53)
we do as design system maintainers that are really well suited for automation or placed somewhere in the pipeline for us to do and then come back and verify and then token automation and token linting and checking and diffing I think is one of those. Yeah, that makes sense. When I looked at these two answers, it's funny because if you just look at the charts, they're almost identical, right?
ep073 speaker raw (06:53)
we do as design system maintainers that are really well suited for automation or placed somewhere in the pipeline for us to do and then come back and verify and then token automation and token linting and checking and diffing I think is one of those. Yeah, that makes sense. When I looked at these two answers, it's funny because if you just look at the charts, they're almost identical, right?
ep073 screen raw (07:22)
Um, and so again, the difference between these questions was just the word could and should. So where could we automate things? Where should we? And, um, you know, we're trying to, we're trying to hone in on like, I don't know the, the, the, the choices that we're making when we choose to sort of like, say, this is a task I am as an individual no longer going to do manually. And what, is the cost of that? When I started to dig in further, what I learned is that about 25%, so about 50 % of you.
Ben Callahan (07:22)
Um, and so again, the difference between these questions was just the word could and should. So where could we automate things? Where should we? And, um, you know, we're trying to, we're trying to hone in on like, I don't know the, the, the, the choices that we're making when we choose to sort of like, say, this is a task I am as an individual no longer going to do manually. And what, is the cost of that? When I started to dig in further, what I learned is that about 25%, so about 50 % of you.
ep073 speaker raw (07:22)
Um, and so again, the difference between these questions was just the word could and should. So where could we automate things? Where should we? And, um, you know, we're trying to, we're trying to hone in on like, I don't know the, the, the, the choices that we're making when we choose to sort of like, say, this is a task I am as an individual no longer going to do manually. And what, is the cost of that? When I started to dig in further, what I learned is that about 25%, so about 50 % of you.
ep073 screen raw (07:52)
gave the exact same answer for both questions. So if you said 50%, you said in the first one, you said that's what it should be. So that's half of us. 25 % decreased their answers between could and should. So they said, we should decrease, essentially we should do less. And 25 % increased, which was like, we should increase the amount of automation. So it's a very interesting split. And so then I started looking across those... ⁓
Ben Callahan (07:52)
gave the exact same answer for both questions. So if you said 50%, you said in the first one, you said that's what it should be. So that's half of us. 25 % decreased their answers between could and should. So they said, we should decrease, essentially we should do less. And 25 % increased, which was like, we should increase the amount of automation. So it's a very interesting split. And so then I started looking across those... ⁓
ep073 speaker raw (07:52)
gave the exact same answer for both questions. So if you said 50%, you said in the first one, you said that's what it should be. So that's half of us. 25 % decreased their answers between could and should. So they said, we should decrease, essentially we should do less. And 25 % increased, which was like, we should increase the amount of automation. So it's a very interesting split. And so then I started looking across those... ⁓
ep073 screen raw (08:19)
data that those data points like who increased who decreased into the way they answered questions three and four. And that's where I started to see some really kind of interesting things emerge in my assessment. One of the most interesting ones. And I think, ⁓ let's see, I think you called out this one too ⁓ as well. Is that right? Davies, is the avoidance reasoning? These two seemed really interesting, right? So the folks who decreased from Q1 to Q2
Ben Callahan (08:19)
data that those data points like who increased who decreased into the way they answered questions three and four. And that's where I started to see some really kind of interesting things emerge in my assessment. One of the most interesting ones. And I think, ⁓ let's see, I think you called out this one too ⁓ as well. Is that right? Davies, is the avoidance reasoning? These two seemed really interesting, right? So the folks who decreased from Q1 to Q2
ep073 speaker raw (08:19)
data that those data points like who increased who decreased into the way they answered questions three and four. And that's where I started to see some really kind of interesting things emerge in my assessment. One of the most interesting ones. And I think, ⁓ let's see, I think you called out this one too ⁓ as well. Is that right? Davies, is the avoidance reasoning? These two seemed really interesting, right? So the folks who decreased from Q1 to Q2
ep073 screen raw (08:50)
They use the word the phrase human in the loop to mean sort of humans are essential to the process and to the learning The group who increased kind of used things that that phrase human in the loop more to mean like hey We just need a final check like it's somewhere at the end We need to give it a thumbs up. And so that's actually a very different Mental model for how humans are involved in the work, right? So that was one thing I thought was really interesting and then David you want to share a little bit about what you saw with this one?
Ben Callahan (08:50)
They use the word the phrase human in the loop to mean sort of humans are essential to the process and to the learning The group who increased kind of used things that that phrase human in the loop more to mean like hey We just need a final check like it's somewhere at the end We need to give it a thumbs up. And so that's actually a very different Mental model for how humans are involved in the work, right? So that was one thing I thought was really interesting and then David you want to share a little bit about what you saw with this one?
ep073 speaker raw (08:50)
They use the word the phrase human in the loop to mean sort of humans are essential to the process and to the learning The group who increased kind of used things that that phrase human in the loop more to mean like hey We just need a final check like it's somewhere at the end We need to give it a thumbs up. And so that's actually a very different Mental model for how humans are involved in the work, right? So that was one thing I thought was really interesting and then David you want to share a little bit about what you saw with this one?
ep073 screen raw (09:18)
yeah, think the in terms of like avoidance reasoning and avoidance is also like I think a strong word like we've seen on my on my team that there's folks that are really interested in things like building components, ⁓ maintaining components, generating new components, generating instances, all that stuff. And they because that's like a part of the work that they deeply enjoy, they do not want to automate.
Ben Callahan (09:18)
yeah, think the in terms of like avoidance reasoning and avoidance is also like I think a strong word like we've seen on my on my team that there's folks that are really interested in things like building components, ⁓ maintaining components, generating new components, generating instances, all that stuff. And they because that's like a part of the work that they deeply enjoy, they do not want to automate.
ep073 speaker raw (09:18)
yeah, think the in terms of like avoidance reasoning and avoidance is also like I think a strong word like we've seen on my on my team that there's folks that are really interested in things like building components, ⁓ maintaining components, generating new components, generating instances, all that stuff. And they because that's like a part of the work that they deeply enjoy, they do not want to automate.
ep073 screen raw (09:46)
any of that stuff. So even things like going in and doing like manual review, they enjoy the that sort of stuff. And they enjoy that. I call it like low level work, right? Much like the folks that love to maintain tokens and like to manually curate them. So it's more of this. That's the way they like to work. And they're not necessarily tuned to modify that thinking quite yet. Yeah.
Ben Callahan (09:46)
any of that stuff. So even things like going in and doing like manual review, they enjoy the that sort of stuff. And they enjoy that. I call it like low level work, right? Much like the folks that love to maintain tokens and like to manually curate them. So it's more of this. That's the way they like to work. And they're not necessarily tuned to modify that thinking quite yet. Yeah.
ep073 speaker raw (09:46)
any of that stuff. So even things like going in and doing like manual review, they enjoy the that sort of stuff. And they enjoy that. I call it like low level work, right? Much like the folks that love to maintain tokens and like to manually curate them. So it's more of this. That's the way they like to work. And they're not necessarily tuned to modify that thinking quite yet. Yeah.
ep073 screen raw (10:17)
Very cool. So lots of interesting stuff in the data folks would love to hear your assessment. And this is the moment where ⁓ we open it up. So if you have questions, if you have ideas, you can raise your hand in zoom. We will certainly call on you. Kaelig first to first to it, buddy. What you got? Hey, thanks for the for setting this up today. Yeah, quick question. So do you think this is how people like to work or is this how this is the
Ben Callahan (10:17)
Very cool. So lots of interesting stuff in the data folks would love to hear your assessment. And this is the moment where ⁓ we open it up. So if you have questions, if you have ideas, you can raise your hand in zoom. We will certainly call on you. Kaelig first to first to it, buddy. What you got? Hey, thanks for the for setting this up today. Yeah, quick question. So do you think this is how people like to work or is this how this is the
ep073 speaker raw (10:17)
Very cool. So lots of interesting stuff in the data folks would love to hear your assessment. And this is the moment where ⁓ we open it up. So if you have questions, if you have ideas, you can raise your hand in zoom. We will certainly call on you. Kaelig first to first to it, buddy. What you got? Hey, thanks for the for setting this up today. Yeah, quick question. So do you think this is how people like to work or is this how this is the
ep073 screen raw (10:44)
only way they know how to work and therefore there's a lack of know-how and automation is not really in ⁓ their DNA and that's something they would need to learn. What do mean by this? you saying like using AI automate things? Yeah, mean using AI or just automating period because I can't think of somebody who would deliberately want to
Ben Callahan (10:44)
only way they know how to work and therefore there's a lack of know-how and automation is not really in ⁓ their DNA and that's something they would need to learn. What do mean by this? you saying like using AI automate things? Yeah, mean using AI or just automating period because I can't think of somebody who would deliberately want to
ep073 speaker raw (10:44)
only way they know how to work and therefore there's a lack of know-how and automation is not really in their DNA and that's something they would need to learn. What do mean by this? you saying like using AI or automating things? Yeah, mean using AI or just automating period because I can't think of somebody who would deliberately want to
ep073 screen raw (11:12)
do these things manually, all the syncing between all these sources, exporting manually, syncing manually. If somebody knew how to automate that, wouldn't they just do it? David, do you want to respond to that? If somebody knew how to automate, wouldn't they just do it?
Ben Callahan (11:12)
do these things manually, all the syncing between all these sources, exporting manually, syncing manually. If somebody knew how to automate that, wouldn't they just do it? Davy do you want to respond to that? If somebody knew how to automate, wouldn't they just do it?
ep073 speaker raw (11:12)
do these things manually, all the syncing between all these sources, exporting manually, syncing manually. If somebody knew how to automate that, wouldn't they just do it? David, do you want to respond to that? If somebody knew how to automate, wouldn't they just do it?
Ben Callahan (11:33)
Yes, yes, and I think part of it is ⁓ wanting, like, ⁓ leaving, ⁓ having like control like in your, in your hand. So that might be like, ⁓ potentially like a job security sort of thing, or like an ownership sort of thing. Like I want to be able to either control the syncing or even if you do have automation, potentially being like the only one that is able to do it. So. ⁓
ep073 screen raw (11:33)
Yes, yes, and I think part of it is ⁓ wanting, like, ⁓ leaving, ⁓ having like control like in your, in your hand. So that might be like, ⁓ potentially like a job security sort of thing, or like an ownership sort of thing. Like I want to be able to either control the syncing or even if you do have automation, potentially being like the only one that is able to do it. So. ⁓
ep073 speaker raw (11:33)
Yes, yes, and I think part of it is ⁓ wanting, like, ⁓ leaving, ⁓ having like control like in your, in your hand. So that might be like, ⁓ potentially like a job security sort of thing, or like an ownership sort of thing. Like I want to be able to either control the syncing or even if you do have automation, potentially being like the only one that is able to do it. So. ⁓
ep073 screen raw (11:59)
I think there's a lot of factors involved, but yeah, that's sort of like where my thinking was going. Yeah, I was sort of, I think for me, like nothing is free in my mind, right? And I think we think of automation as like a way to get something for nothing. And I don't think that that's true. Like, and even these mundane tasks that are a pain in the butt to do and you know,
Ben Callahan (11:59)
I think there's a lot of factors involved, but yeah, that's sort of like where my thinking was going. Yeah, I was sort of, I think for me, like nothing is free in my mind, right? And I think we think of automation as like a way to get something for nothing. And I don't think that that's true. Like, and even these mundane tasks that are a pain in the butt to do and you know,
ep073 speaker raw (11:59)
I think there's a lot of factors involved, but yeah, that's sort of like where my thinking was going. Yeah, I was sort of, I think for me, nothing is free in my mind, right? And I think we think of automation as like a way to get something for nothing. And I don't think that that's true. Like, and even these mundane tasks that are a pain in the butt to do and you know,
ep073 screen raw (12:25)
the syncing, all of that stuff that's very manual, time consuming stuff. Doing it also increases your understanding and there are things you learn in that process. And so I'm not saying we have to always do it, but I do think we have to acknowledge the cost, right? So what are we missing? What are we paying when we get this sort of free automation? Kailik, does that answer your question?
Ben Callahan (12:25)
the syncing, all of that stuff that's very manual, time consuming stuff. Doing it also increases your understanding and there are things you learn in that process. And so I'm not saying we have to always do it, but I do think we have to acknowledge the cost, right? So what are we missing? What are we paying when we get this sort of free automation? Kaelig does that answer your question?
ep073 speaker raw (12:25)
the syncing, all of that stuff that's very manual, time consuming stuff. Doing it also increases your understanding and there are things you learn in that process. And so I'm not saying we have to always do it, but I do think we have to acknowledge the cost, right? So what are we missing? What are we paying when we get this sort of free automation? Kailik, does that answer your question?
Ben Callahan (12:48)
Yeah, yeah, absolutely. And also you're kind of starting to touch on the cost of maintaining those automations, because anytime you put an abstraction in the the path to delivery, this needs to this has a life of its own. So it's not free, as in, like, you just set it up and forget about it. There always are little things that will break or things that need human attention. And it's an interesting balancing act sometimes. Yeah.
ep073 speaker raw (12:48)
Yeah, yeah, absolutely. And also you're kind of starting to touch on the cost of maintaining those automations, because anytime you put an abstraction in the path to delivery, this has a life of its own. So it's not free, as in, like, you just set it up and forget about it. There always are little things that will break or things that need human attention. And it's an interesting balancing act sometimes. Yeah.
ep073 screen raw (12:48)
Yeah, yeah, absolutely. And also you're kind of starting to touch on the cost of maintaining those automations, because anytime you put an abstraction in the the path to delivery, this needs to this has a life of its own. So it's not free, as in, like, you just set it up and forget about it. There always are little things that will break or things that need human attention. And it's an interesting balancing act sometimes. Yeah.
Thank you. Peter, what's up?
Ben Callahan (13:18)
Thank you. Peter, what's up?
ep073 speaker raw (13:18)
Thank you. Peter, what's up?
ep073 screen raw (13:22)
Yeah, I I think the thing that I wanted to kind of bring up is, and it was something else that some other folks brought up in the chat as well, was I think we need to be kind of clear about automation versus AI and be very clear about those definitions because like the way that I've been working, which is heavily, heavily into AI, I do not necessarily consider automation. Like I consider like working with AI fairly manual in a sense, even if it is performing actions.
Ben Callahan (13:22)
Yeah, I I think the thing that I wanted to kind of bring up is, and it was something else that some other folks brought up in the chat as well, was I think we need to be kind of clear about automation versus AI and be very clear about those definitions because like the way that I've been working, which is heavily, heavily into AI, I do not necessarily consider automation. Like I consider like working with AI fairly manual in a sense, even if it is performing actions.
ep073 speaker raw (13:22)
Yeah, I I think the thing that I wanted to kind of bring up is, and it was something else that some other folks brought up in the chat as well, was I think we need to be kind of clear about automation versus AI and be very clear about those definitions because like the way that I've been working, which is heavily, heavily into AI, I do not necessarily consider automation. Like I consider like working with AI fairly manual in a sense, even if it is performing actions.
ep073 screen raw (13:50)
But automated, would define as basically kind of workflows that are coded or are kind of process enabled that just run automatically without really kind of any kind of, you know.
Ben Callahan (13:50)
But automated, would define as basically kind of workflows that are coded or are kind of process enabled that just run automatically without really kind of any kind of, you know.
ep073 speaker raw (13:50)
But automated, would define as basically kind of workflows that are coded or are kind of process enabled that just run automatically without really kind of any kind of, you know.
ep073 screen raw (14:06)
of thought or LLM into it. So for example, like the way my tokens are created, kind of suit it. Like I've worked with AI to kind of create and add the tokens, but then I have an automated flow to kind of create the tokens for iOS, Android and web ⁓ in basically a ⁓
Ben Callahan (14:06)
of thought or LLM into it. So for example, like the way my tokens are created, kind of suit it. Like I've worked with AI to kind of create and add the tokens, but then I have an automated flow to kind of create the tokens for iOS, Android and web ⁓ in basically a ⁓
ep073 speaker raw (14:06)
of thought or LLM into it. So for example, like the way my tokens are created, kind of suit it. Like I've worked with AI to kind of create and add the tokens, but then I have an automated flow to kind of create the tokens for iOS, Android and web ⁓ in basically a ⁓
Ben Callahan (14:29)
Yeah, it's just basically just run code and then make things go out is more the way that I think about automation. And that's actually, that's an area that I kind of got caught up a little bit in the survey was kind of like flip-flopping between those two kind of definitions. So I think it actually even shouldn't kind of be have an idea, like how does my definition of automation AI compare to like everybody else's? Yeah, yeah, it's a good point. I think that the intent here with this was to talk about using AI to do automation.
ep073 screen raw (14:29)
Yeah, it's just basically just run code and then make things go out is more the way that I think about automation. And that's actually, that's an area that I kind of got caught up a little bit in the survey was kind of like flip-flopping between those two kind of definitions. So I think it actually even shouldn't kind of be, have an idea, like how does my definition of automation AI compare to like everybody else's? Yeah, yeah, it's a good point. I think that the intent here with this was to talk about using AI to do automation.
ep073 speaker raw (14:29)
Yeah, it's just basically just run code and then make things go out is more the way that I think about automation. And that's actually, that's an area that I kind of got caught up a little bit in the survey was kind of like flip-flopping between those two kind of definitions. So I think even shouldn't kind of be, have an idea, like how does my definition of automation AI compare to like everybody else's? Yeah, yeah, it's a good point. I think that the intent here with this was to talk about using AI to do automation.
ep073 screen raw (14:57)
So it's the overlap of those two things and not, we weren't intending to conflate them, right? Like, of course there are automations that don't involve AI, there have been for decades at this point. ⁓ So we're not trying to say they're the same thing. So great call out. ⁓ I don't know if others had that same kind of confusion in the questions or not, but Dave, anything you wanna add to that or?
Ben Callahan (14:57)
So it's the overlap of those two things and not, we weren't intending to conflate them, right? Like, of course there are automations that don't involve AI, there have been for decades at this point. ⁓ So we're not trying to say they're the same thing. So great call out. ⁓ I don't know if others had that same kind of confusion in the questions or not, but Dave, anything you wanna add to that or?
ep073 speaker raw (14:57)
So it's the overlap of those two things and not, we weren't intending to conflate them, right? Like, of course there are automations that don't involve AI, there have been for decades at this point. ⁓ So we're not trying to say they're the same thing. So great call out. ⁓ I don't know if others had that same kind of confusion in the questions or not, but Dave, anything you wanna add to that or?
ep073 screen raw (15:20)
No, that's about it. And Sarah pointed out that to you. The intersection of the two. one example might be, now we as designers, design system maintainers, we have the ability to utilize AI to write these simple automations that run on their own, but we do need to watch the watchers that are doing the work from time to time, obviously. I like that. Doug, what's up?
Ben Callahan (15:20)
No, that's about it. And Sarah pointed out that to you. The intersection of the two. ⁓ One example might be, we as designers, design system maintainers, ⁓ we have the ability to utilize AI to write these simple automations that run on their own, but we do need to watch the ⁓ watchers that ⁓ are doing the work from time to time, obviously. ⁓
ep073 speaker raw (15:20)
No, that's about it. And Sarah pointed out that to you. The intersection of the two. ⁓ One example might be, we as designers, design system maintainers, ⁓ we have the ability to utilize AI to write these simple automations that run on their own, but we do need to watch the ⁓ watchers that ⁓ are doing the work from time to time, obviously. I like that. Doug, what's up?
ep073 screen raw (15:51)
We had an interesting thing. I definitely tend towards deterministic code. So dealing with the non deterministic nature of AI is something I don't reach for for automation as much as I probably could. But on our team, have our release process is somewhat automated, but then there's stuff that we do just because we care about the output and we've not bothered to completely fix our release process, our automated code release process. So there's different things that we check and if it interests this thing or if it incorrectly picks up this change, we back it out.
Ben Callahan (15:51)
We had an interesting thing. I definitely tend towards deterministic code. So dealing with the non deterministic nature of AI is something I don't reach for for automation as much as I probably could. But on our team we have our release process is somewhat automated, but then there's stuff that we do just because we care about the output and we've not bothered to completely fix our release process, our automated code release process. So there's different things that we check and if it interests this thing or if it incorrectly picks up this change, we back it out.
ep073 speaker raw (15:51)
We had an interesting thing. I definitely tend towards deterministic code. So dealing with the non deterministic nature of AI is something I don't reach for for automation as much as I probably could. But on our team we have our release process is somewhat automated, but then there's stuff that we do just because we care about the output and we've not bothered to completely fix our release process, our automated code release process. So there's different things that we check and if it interests this thing or if it incorrectly picks up this change, we back it out.
ep073 screen raw (16:19)
Well, my coworker turned the whole thing into a skill. So he has AI do the releases now and he asks questions and does stuff, it, instead of spending a couple hours figuring out the tooling and changing it so we don't have to have those steps, AI is smoothing over the thing. And so he was able to do that really quickly and it's been very consistent. So it's just a different way of thinking about it. I would, I would spend time trying to solve the automation. it's deterministic every single time. And for him, the non-deterministic nature allowed it to be quite flexible without
Ben Callahan (16:19)
Well, my coworker turned the whole thing into a skill. So he has AI do the releases now and he asks questions and does stuff, it, instead of spending a couple hours figuring out the tooling and changing it so we don't have to have those steps, AI is smoothing over the thing. And so he was able to do that really quickly and it's been very consistent. So it's just a different way of thinking about it. I would, I would spend time trying to solve the automation. it's deterministic every single time. And for him, the non-deterministic nature allowed it to be quite flexible without
ep073 speaker raw (16:19)
Well, my coworker turned the whole thing into a skill. So he has AI do the releases now and he asks questions and does stuff, it, instead of spending a couple hours figuring out the tooling and changing it so we don't have to have those steps, AI is smoothing over the thing. And so he was able to do that really quickly and it's been very consistent. So it's just a different way of thinking about it. I would, I would spend time trying to solve the automation. it's deterministic every single time. And for him, the non-deterministic nature allowed it to be quite flexible without
ep073 screen raw (16:49)
changing our tooling. The tooling is exactly the same as we using before. So that was interesting. I like that. Randy. Hi, long time listener, first time caller. So I just wanted to bring up that automation is really in the life cycle of a process. It is the mature end of a process. Like automation is not in and of itself just a goal. It's running something that's proven.
Ben Callahan (16:49)
changing our tooling. The tooling is exactly the same as we using before. So that was interesting. I like that. Randy. Hi, long time listener, first time caller. So I just wanted to bring up that automation is really in the life cycle of a process. It is the mature end of a process. Like automation is not in and of itself just a goal. It's running something that's proven.
ep073 speaker raw (16:49)
changing our tooling. The tooling is exactly the same as we were using before. So that was interesting. I like that. Randy. Hi, long time listener, first time caller. So I just wanted to bring up that automation is really in the life cycle of a process. It is the mature end of a process. Like automation is not in and of itself just a goal. It's running something that's proven.
ep073 screen raw (17:18)
Like, you know, kind of like building off of what Douglas said, you know, you want to be able to have reproducible results. And so like jumping straight to automation may, may be successful because maybe AI understands what you're trying to do, but it is better to think of in terms of like automation as the, like the final form of a cow path, you know, you've got like a path that people take, and then you start like putting down paving stones. And then suddenly it turns into a road years later.
Ben Callahan (17:18)
Like, you know, kind of like building off of what Douglas said, you know, you want to be able to have reproducible results. And so like jumping straight to automation may, may be successful because maybe AI understands what you're trying to do, but it is better to think of in terms of like automation as the, like the final form of a cow path, you know, you've got like a path that people take, and then you start like putting down paving stones. And then suddenly it turns into a road years later.
ep073 speaker raw (17:18)
Like, you know, kind of like building off of what Douglas said, you know, you want to be able to have reproducible results. And so like jumping straight to automation may, may be successful because maybe AI understands what you're trying to do, but it is better to think of in terms of like automation as the, like the final form of a cow path, you know, you've got like a path that people take, and then you start like putting down paving stones. And then suddenly it turns into a road years later.
ep073 screen raw (17:48)
you know, it's automation works best whenever it is a process that, has proven. And that's something that I think like a lot of my experiments with AI, like when I try to jump past that thinking through the process and having it prove, ⁓ have it become proven, ⁓ is where I run into the most problems with AI. Yeah, I like that actually. And I think that what's cool about that is it allows us
Ben Callahan (17:48)
you know, it's automation works best whenever it is a process that, has proven. And that's something that I think like a lot of my experiments with AI, like when I try to jump past that thinking through the process and having it prove, ⁓ have it become proven, ⁓ is where I run into the most problems with AI. Yeah, I like that actually. And I think that what's cool about that is it allows us
ep073 speaker raw (17:48)
you know, it's automation works best whenever it is a process that, has proven. And that's something that I think like a lot of my experiments with AI, like when I try to jump past that thinking through the process and having it prove, ⁓ have it become proven, ⁓ is where I run into the most problems with AI. Yeah, I like that actually. And I think that what's cool about that is it allows us
ep073 screen raw (18:16)
some of the learning, know, still as humans, right, to sort of like work through it, get into the weeds, do the manual migration or whatever it is, so that we know what to automate, you know, so that's I think that's a healthy way of thinking about it. Davie, any thoughts on that? No, I think the maturity also ties back to like this a bit. I do want to mention back to what Douglas was saying is like a mechanism to do this as a repeatable fashion is through through skills. So
Ben Callahan (18:16)
some of the learning, know, still as humans, right, to sort of like work through it, get into the weeds, do the manual migration or whatever it is, so that we know what to automate, you know, so that's I think that's a healthy way of thinking about it. Davie, any thoughts on that? No, I think the maturity also ties back to like this a bit. I do want to mention back to what Douglas was saying is like a mechanism to do this as a repeatable fashion is through through skills. So
ep073 speaker raw (18:16)
some of the learning, know, still as humans, right, to sort of like work through it, get into the weeds, do the manual migration or whatever it is, so that we know what to automate, you know, so that's I think that's a healthy way of thinking about it. Davie, any thoughts on that? No, I think the maturity also ties back to like this a bit. I do want to mention back to what Douglas was saying is like a mechanism to do this as a repeatable fashion is through through skills. So
ep073 screen raw (18:45)
whether you have a process that works or if you are not yet at that maturity level, but you have another team that has captured this as a skill, that's another mechanism to try it out. Yeah, I love it. Sarah, jump in. Yeah, I love the last two comments, kind of ⁓ talking about like the maturity of your design system and using AI for automation.
Ben Callahan (18:45)
whether you have a process that works or if you are not yet at that maturity level, but you have another team that has captured this as a skill, that's another mechanism to try it out. Yeah, I love it. Sarah, jump in. Yeah, I love the last two comments, kind of ⁓ talking about like the maturity of your design system and using AI for automation.
ep073 speaker raw (18:45)
whether you have a process that works or if you are not yet at that maturity level, but you have another team that has captured this as a skill, that's another mechanism to try it out. Yeah, I love it. Sarah, jump in. Yeah, I love the last two comments, kind of ⁓ talking about like the maturity of your design system and using AI for automation.
ep073 screen raw (19:11)
Um, something that Davy had said earlier, talking about the avoidance is really, it's like changing your mind shift. And I've been trying to do that with my team is really, it's a mindset shift and AI is a tool. And I feel like we so much are like, solve everything with AI. But, um, I think somebody had mentioned earlier, it's like looking at the actual problem and what are we trying to solve? Um, whereas like it's, it is that, um, the person before me, sorry.
ep073 speaker raw (19:11)
Um, something that Davy had said earlier, talking about the avoidance is really, it's like changing your mind shift. And I've been trying to do that with my team is really, it's a mindset shift and AI is a tool. And I feel like we so much are like, solve everything with AI. But, um, I think somebody had mentioned earlier, it's like looking at the actual problem and what are we trying to solve? Um, whereas like it's, it is that, um, the person before me, sorry.
Ben Callahan (19:11)
something that Davy had said earlier, talking about the avoidance is really, it's like changing your mind shift. And I've been trying to do that with my team is really, it's a mindset shift and AI is a tool. And I feel like we so much are like, solve everything with AI. But, I think somebody had mentioned earlier, it's like looking at the actual problem and what are we trying to solve? ⁓ whereas like it's, it is that, ⁓ the person before me, sorry.
ep073 screen raw (19:40)
forgot your name already, was talking about the, like, you have to like have that process and then look at the whole process and then see where automation fits in. But I think it's the same as like, where does AI fit in to all of that? And it's a mindset shift, right? It's like starting with a where can AI fit in and where does it best fit in sort of back to the original question, what could be automated versus what should be automated? I feel like we should be asking those questions
Ben Callahan (19:40)
forgot your name already, was talking about the, like, you have to like have that process and then look at the whole process and then see where automation fits in. But I think it's the same as like, where does AI fit in to all of that? And it's a mindset shift, right? It's like starting with a where can AI fit in and where does it best fit in sort of back to the original question, what could be automated versus what should be automated? I feel like we should be asking those questions
ep073 speaker raw (19:40)
forgot your name already, was talking about the, like, you have to like have that process and then look at the whole process and then see where automation fits in. But I think it's the same as like, where does AI fit in to all of that? And it's a mindset shift, right? It's like starting with a where can AI fit in and where does it best fit in sort of back to the original question, what could be automated versus what should be automated? I feel like we should be asking those questions
ep073 screen raw (20:10)
at every turn and shifting our mindset toward what is the best implementation of AI to solve these problems. Sarah, are you feeling ⁓ like it? I love that sentiment that it's we need to be much more intentional about the places where we do this. Are you feeling pressure organizationally to not have that consideration? Like what's that feels to me like that would be table stakes for this kind of thing?
Ben Callahan (20:10)
at every turn and shifting our mindset toward what is the best implementation of AI to solve these problems. Sarah, are you feeling ⁓ like it? I love that sentiment that it's we need to be much more intentional about the places where we do this. Are you feeling pressure organizationally to not have that consideration? Like what's that feels to me like that would be table stakes for this kind of thing?
ep073 speaker raw (20:10)
at every turn and shifting our mindset toward what is the best implementation of AI to solve these problems. Sarah, are you feeling ⁓ like it? I love that sentiment that it's we need to be much more intentional about the places where we do this. Are you feeling pressure organizationally to not have that consideration? Like what's that feels to me like that would be table stakes for this kind of thing?
ep073 screen raw (20:39)
But I also hear what you're saying, which is like, feels like a lot of people are just automating the heck out of everything because they can. So what's the tension that you say that? Yeah, I would say at Target, I mean, it's like AI running on AI, moving from using AI to running AI every day all day, AI AI, right? Yeah. And so if you can check that box, show your leader, hey, I'm using AI. like, it feels like that's what everybody needs to be doing. And we're at a place at Target where
Ben Callahan (20:39)
But I also hear what you're saying, which is like, feels like a lot of people are just automating the heck out of everything because they can. So what's the tension that you say that? Yeah, I would say at Target, I mean, it's like AI running on AI, moving from using AI to running AI every day all day, AI AI, right? Yeah. And so if you can check that box, show your leader, hey, I'm using AI. like, it feels like that's what everybody needs to be doing. And we're at a place at Target where
ep073 speaker raw (20:39)
But I also hear what you're saying, which is like, feels like a lot of people are just automating the heck out of everything because they can. So what's the tension that you say that? Yeah, I would say at Target, I mean, it's like AI running on AI, moving from using AI to running AI every day all day, AI AI, right? Yeah. And so if you can check that box, show your leader, hey, I'm using AI. like, it feels like that's what everybody needs to be doing. And we're at a place at Target where
ep073 screen raw (21:09)
where there is a shift happening where it is this like everybody's just learning and trying to use AI everywhere they can without the thoughtfulness. And I think there is like this, I call it the pendulum swing, right? It's like everybody's using and it's all over here and it's everywhere. And it's swinging back sort of to the middle where we kind of talked about this ⁓ like a month or so ago is the context and the architecture of AI solutioning. ⁓
Ben Callahan (21:09)
where there is a shift happening where it is this like everybody's just learning and trying to use AI everywhere they can without the thoughtfulness. And I think there is like this, I call it the pendulum swing, right? It's like everybody's using and it's all over here and it's everywhere. And it's swinging back sort of to the middle where we kind of talked about this ⁓ like a month or so ago is the context and the architecture of AI solutioning. ⁓
ep073 speaker raw (21:09)
where there is a shift happening where it is this like everybody's just learning and trying to use AI everywhere they can without the thoughtfulness. And I think there is like this, I call it the pendulum swing, right? It's like everybody's using and it's all over here and it's everywhere. And it's swinging back sort of to the middle where we kind of talked about this ⁓ like a month or so ago is the context and the architecture of AI solutioning. ⁓
ep073 screen raw (21:37)
really needs to be thought through before we're building a bunch of stuff. Like people just jump straight to agents without thinking about in a design system world, you need to have like solid context to provide to large language models to work within. So yeah, I think there is definitely a push for that. But in doing so at a large organization like Target, we're also seeing a lot of overlap where people are building the same things and solving some of the same problems.
Ben Callahan (21:37)
really needs to be thought through before we're building a bunch of stuff. Like people just jump straight to agents without thinking about in a design system world, you need to have like solid context to provide to large language models to work within. So yeah, I think there is definitely a push for that. But in doing so at a large organization like Target, we're also seeing a lot of overlap where people are building the same things and solving some of the same problems.
ep073 speaker raw (21:37)
really needs to be thought through before we're building a bunch of stuff. Like people just jump straight to agents without thinking about in a design system world, you need to have like solid context to provide to large language models to work within. So yeah, I think there is definitely a push for that. But in doing so at a large organization like Target, we're also seeing a lot of overlap where people are building the same things and solving some of the same problems.
ep073 screen raw (22:07)
without talking to each other. some of it is reusable and some of it is like shareable and other stuff. It's like, you didn't think about these 10 contexts because that's not really your domain. You're just trying to like jump straight to a solution. Yeah. So good. Thanks Sarah for the extra context. Davy, I'm going to put you on the spot here because when we first looked at this, you were like,
Ben Callahan (22:07)
without talking to each other. some of it is reusable and some of it is like shareable and other stuff. It's like, you didn't think about these 10 contexts because that's not really your domain. You're just trying to like jump straight to a solution. yeah. Yeah. So good. Thanks Sarah for the extra context. Davy, I'm going to put you on the spot here because when we first looked at this, you were like,
ep073 speaker raw (22:07)
without talking to each other. some of it is reusable and some of it is like shareable and other stuff. It's like, you didn't think about these 10 contexts because that's not really your domain. You're just trying to like jump straight to a solution. Yeah. So good. Thanks Sarah for the extra context. Davy, I'm going to put you on the spot here because when we first looked at this, you were like,
ep073 screen raw (22:35)
I'm nervous because I'm trying to automate everything. So I want you to respond to Sarah. Like how are you feeling about all this right now? Yeah, I think there's like the the step where you're you get all these different tasks or these asks from your your design leads and your managers and such. And then you step back, take a deep breath. Like how much of this stuff do I like to do? How much do I want to do? How much of this can I automate? And then how much of this stuff can I
Ben Callahan (22:35)
I'm nervous because I'm trying to automate everything. So I want you to respond to Sarah. Like how are you feeling about all this right now? Yeah, I think there's like the the step where you're you get all these different tasks or these asks from your your design leads and your managers and such. And then you step back, take a deep breath. Like how much of this stuff do I like to do? How much do I want to do? How much of this can I automate? And then how much of this stuff can I
ep073 speaker raw (22:35)
I'm nervous because I'm trying to automate everything. So I want you to respond to Sarah. Like how are you feeling about all this right now? Yeah, I think there's like the the step where you're you get all these different tasks or these asks from your your design leads and your managers and such. And then you step back, take a deep breath. Like how much of this stuff do I like to do? How much do I want to do? How much of this can I automate? And then how much of this stuff can I
ep073 screen raw (23:05)
mentally run in parallel. So that's like another aspect of this, which is, I think, especially for like, ⁓ folks that have a hard time like context switching, like running, being able to run a lot of these jobs, and potentially maybe these tasks are research tasks or discovery tasks, running them in parallel while you're working on another thing. yeah, I'm of the mind, I think, where I'm trying to be at like the 60.
Ben Callahan (23:05)
mentally run in parallel. So that's like another aspect of this, which is, I think, especially for like, ⁓ folks that have a hard time like context switching, like running, being able to run a lot of these jobs, and potentially maybe these tasks are research tasks or discovery tasks, running them in parallel while you're working on another thing. yeah, I'm of the mind, I think, where I'm trying to be at like the 60.
ep073 speaker raw (23:05)
mentally run in parallel. So that's like another aspect of this, which is, I think, especially for like, ⁓ folks that have a hard time like context switching, like running, being able to run a lot of these jobs, and potentially maybe these tasks are research tasks or discovery tasks, running them in parallel while you're working on another thing. yeah, I'm of the mind, I think, where I'm trying to be at like the 60.
ep073 screen raw (23:34)
ish percent 70 % of my work is starting off with some sort of automation and AI to support myself. Yeah. Interesting. And Bill, I want to bring you in my friend, I think you I'm also going to call you out. You said 0 % and 0 % I think in your answers to questions one and two. So give it to us, buddy. I'm not going to be too, too down on this. And I think I would change that actually thinking about what Doug said about
Ben Callahan (23:34)
ish percent 70 % of my work is starting off with some sort of automation and AI to support myself. Yeah. Interesting. And Bill, I want to bring you in my friend, I think you I'm also going to call you out. You said 0 % and 0 % I think in your answers to questions one and two. So give it to us, buddy. I'm not going to be too, too down on this. And I think I would change that actually thinking about what Doug said about
ep073 speaker raw (23:34)
ish percent 70 % of my work is starting off with some sort of automation and AI to support myself. Yeah. Interesting. And Bill, I want to bring you in my friend, I think you I'm also going to call you out. You said 0 % and 0 % I think in your answers to questions one and two. So give it to us, buddy. I'm not going to be too, too down on this. And I think I would change that actually thinking about what Doug said about
ep073 screen raw (24:03)
⁓ Having just a little co-pilot there on your release automation is very useful. ⁓ And people have talked about the determinism and how critical that is. And there's areas where determinism is just too important. I I used to be a sort of principally a build engineer and having to worry about what the random seed on the OS was, was critical to reproducible build. So you can imagine like putting some matrix multiplication into that is a recipe for disaster.
Ben Callahan (24:03)
⁓ Having just a little co-pilot there on your release automation is very useful. ⁓ And people have talked about the determinism and how critical that is. And there's areas where determinism is just too important. I I used to be a sort of principally a build engineer and having to worry about what the random seed on the OS was, was critical to reproducible build. So you can imagine like putting some matrix multiplication into that is a recipe for disaster.
ep073 speaker raw (24:03)
⁓ Having just a little co-pilot there on your release automation is very useful. ⁓ And people have talked about the determinism and how critical that is. And there's areas where determinism is just too important. I I used to be a sort of principally a build engineer and having to worry about what the random seed on the OS was, was critical to reproducible build. So you can imagine like putting some matrix multiplication into that is a recipe for disaster.
ep073 screen raw (24:31)
But I think the more interesting thing is what the last few people have been talking about ⁓ in terms of at what point you go for AI. And I think one of the risks with using AI anywhere ⁓ is that it doesn't challenge you enough necessarily. And lot of automation comes out of human boredom and frustration with having to do a repetitive task. Computers don't mind doing that. We do.
Ben Callahan (24:31)
But I think the more interesting thing is what the last few people have been talking about ⁓ in terms of at what point you go for AI. And I think one of the risks with using AI anywhere ⁓ is that it doesn't challenge you enough necessarily. And lot of automation comes out of human boredom and frustration with having to do a repetitive task. Computers don't mind doing that. We do.
ep073 speaker raw (24:31)
But I think the more interesting thing is what the last few people have been talking about ⁓ in terms of at what point you go for AI. And I think one of the risks with using AI anywhere ⁓ is that it doesn't challenge you enough necessarily. And lot of automation comes out of human boredom and frustration with having to do a repetitive task. Computers don't mind doing that. We do.
ep073 screen raw (24:58)
So we have to try and automate it and having automated a fair number myself, there's somewhere you get to that point where you go, this is really hard to automate, apps, it's not a good process. Like by using the enormous power of AI to allow you to automate away any mundane task or difficult task, you lose the opportunity to say, well, does it need to be this difficult? Are we actually doing the right thing? Which I know, it's not to say that you can't use AI to automate it in the end, but it's...
Ben Callahan (24:58)
So we have to try and automate it and having automated a fair number myself, there's somewhere you get to that point where you go, this is really hard to automate, apps, it's not a good process. Like by using the enormous power of AI to allow you to automate away any mundane task or difficult task, you lose the opportunity to say, well, does it need to be this difficult? Are we actually doing the right thing? Which I know, it's not to say that you can't use AI to automate it in the end, but it's...
ep073 speaker raw (24:58)
So we have to try and automate it and having automated a fair number myself, there's somewhere you get to that point where you go, this is really hard to automate, apps, it's not a good process. Like by using the enormous power of AI to allow you to automate away any mundane task or difficult task, you lose the opportunity to say, well, does it need to be this difficult? Are we actually doing the right thing? Which I know, it's not to say that you can't use AI to automate it in the end, but it's...
Ben Callahan (25:27)
It's just one of those things where AI doesn't really always challenge you when you say, hey, automate this. And it doesn't say why. That's a stupid thing to do. Whereas if you go and speak to your release engineering manager team people, and they say, no, we can't do that because it's, you you just lose that human conversation and discovery of what is actually a good thing to do. Yeah. Okay. So to summarize that, Bill, you're saying in previous time before AI,
ep073 speaker raw (25:27)
It's just one of those things where AI doesn't really always challenge you when you say, hey, automate this. And it doesn't say why. That's a stupid thing to do. Whereas if you go and speak to your release engineering manager team people, and they say, no, we can't do that because it's, you you just lose that human conversation and discovery of what is actually a good thing to do. Yeah. Okay. So to summarize that, Bill, you're saying in previous time before AI,
ep073 screen raw (25:27)
It's just one of those things where AI doesn't really always challenge you when you say, hey, automate this. And it doesn't say why. That's a stupid thing to do. Whereas if you go and speak to your release engineering manager team people, and they say, no, we can't do that because it's, you you just lose that human conversation and discovery of what is actually a good thing to do. Yeah. Okay. So to summarize that, Bill, you're saying in previous time before AI,
when you would automate things, occasionally you stumbled upon something that in the midst of trying to automate it, you were like, okay, why the heck are we doing things this way? It made you question your process. you, because AI will do it no matter what, you lose that knowledge. Is that kind of the sentiment? Yeah, yeah. It's trying to automate process exposes their flaws and non-determinism in them. so yeah, using AI to solve complex problems is good, but using it to solve non-deterministic problems is.
Ben Callahan (25:56)
when you would automate things, occasionally you stumbled upon something that in the midst of trying to automate it, you were like, okay, why the heck are we doing things this way? It made you question your process. you, because AI will do it no matter what, you lose that knowledge. Is that kind of the sentiment? Yeah, yeah. It's trying to automate process exposes their flaws and non-determinism in them. so yeah, using AI to solve complex problems is good, but using it to solve non-deterministic problems is.
ep073 speaker raw (25:56)
when you would automate things, occasionally you stumbled upon something that in the midst of trying to automate it, you were like, okay, why the heck are we doing things this way? It made you question your process. you, because AI will do it no matter what, you lose that knowledge. Is that kind of the sentiment? Yeah, yeah. It's trying to automate process exposes their flaws and non-determinism in them. so yeah, using AI to solve complex problems is good, but using it to solve non-deterministic problems is.
ep073 screen raw (26:27)
Yeah, very cool. Thanks, Bill. Stephen?
Ben Callahan (26:27)
Yeah, very cool. Thanks, Bill. Stephen?
ep073 speaker raw (26:27)
Yeah, very cool. Thanks, Bill. Stephen?
ep073 screen raw (26:33)
Yeah, I mean, everybody's had such good insights here. So I've had a lot of thoughts, but I'm just going to focus on like a couple things. think the way I've been looking at this is like, see this and I've been pretty much heads down in the AI design system stuff now for the last few months, because that's been the directive I've been given. But like, kind of see it as, I think there is a very clear line with things that are good for AI to do and things that AI are not supposed to do.
Ben Callahan (26:33)
Yeah, I mean, everybody's had such good insights here. So I've had a lot of thoughts, but I'm just going to focus on like a couple things. think the way I've been looking at this is like, see this and I've been pretty much heads down in the AI design system stuff now for the last few months, because that's been the directive I've been given. But like, kind of see it as, I think there is a very clear line with things that are good for AI to do and things that AI are not supposed to do.
ep073 speaker raw (26:33)
Yeah, I mean, everybody's had such good insights here. So I've had a lot of thoughts, but I'm just going to focus on like a couple things. think the way I've been looking at this is like, see this and I've been pretty much heads down in the AI design system stuff now for the last few months, because that's been the directive I've been given. But like, kind of see it as, I think there is a very clear line with things that are good for AI to do and things that AI are not supposed to do.
ep073 screen raw (27:02)
Like I think it's pretty easy to sometimes ask that question of, you know, is this a yes or no on or off thing that is just very black and white, or is this something that needs some like intelligent reasoning behind it ⁓ that, you know, an LLM might accelerate for me that might take me a whole day of thinking about this problem to come up with, you know, some valid solutions or something like that. So I try to apply that.
Ben Callahan (27:02)
Like I think it's pretty easy to sometimes ask that question of, know, is this a yes or no on or off thing that is just very black and white, or is this something that needs some like intelligent reasoning behind it ⁓ that, you know, an LLM might accelerate for me that might take me a whole day of thinking about this problem to come up with, you know, some valid solutions or something like that. So I try to apply that.
ep073 speaker raw (27:02)
Like I think it's pretty easy to sometimes ask that question of, know, is this a yes or no on or off thing that is just very black and white, or is this something that needs some like intelligent reasoning behind it ⁓ that, you know, an LLM might accelerate for me that might take me a whole day of thinking about this problem to come up with, you know, some valid solutions or something like that. So I try to apply that.
ep073 screen raw (27:31)
to a lot of these decisions of, you cause I see a lot of people using AI for stuff like, you know, make sure my code is not doing this thing. And it's like, that's just a test or like a linting rule or, you know, what it might be. And I know we've been talking about tokens lot and, you know, I try to just combine both of these things into one, right? Like when I look at it, I'm like, okay, the token diff, right? Some changes have been made to tokens. I don't want to use AI to figure out what the diff is.
Ben Callahan (27:31)
to a lot of these decisions of, you cause I see a lot of people using AI for stuff like, you know, make sure my code is not doing this thing. And it's like, that's just a test or like a linting rule or, you know, what it might be. And I know we've been talking about tokens lot and, you know, I try to just combine both of these things into one, right? Like when I look at it, I'm like, okay, the token diff, right? Some changes have been made to tokens. I don't want to use AI to figure out what the diff is.
ep073 speaker raw (27:31)
to a lot of these decisions of, you cause I see a lot of people using AI for stuff like, you know, make sure my code is not doing this thing. And it's like, that's just a test or like a linting rule or, you know, what it might be. And I know we've been talking about tokens lot and, you know, I try to just combine both of these things into one, right? Like when I look at it, I'm like, okay, the token diff, right? Some changes have been made to tokens. I don't want to use AI to figure out what the diff is.
Ben Callahan (28:01)
I might want to use AI to explain that diff of the tokens in our token consumer friendly terminology that might be easy for them to understand. I'm not a technical writer. Like, part, I'm an engineer. I'm not ever going to be able to write stuff in a way that people like to read. That's just not who I am. So I can automate that process of, I use it daily to automate pull requests and Slack messages. And here's the changes that have been made to the system.
ep073 speaker raw (28:01)
I might want to use AI to explain that diff of the tokens in our token consumer friendly terminology that might be easy for them to understand. I'm not a technical writer. Like, part, I'm an engineer. I'm not ever going to be able to write stuff in a way that people like to read. That's just not who I am. So I can automate that process of, I use it daily to automate pull requests and Slack messages. And here's the changes that have been made to the system.
ep073 screen raw (28:01)
I might want to use AI to explain that diff of the tokens in our token consumer friendly terminology that might be easy for them to understand. I'm not a technical writer. Like, part, I'm an engineer. I'm not ever going to be able to write stuff in a way that people like to read. That's just not who I am. So I can automate that process of, I use it daily to automate pull requests and Slack messages. And here's the changes that have been made to the system.
list them out in a nice way. That used to take me like half a day or a whole day to like put those types of things together that I can now automate and say, hey, look at all the changes over the last week and like give me a nice executive summary for my boss who wants like three bullet points that are one sentence each. That used to take me so long. And I can automate those things now that allow me to work on the things that are important. Because I really agree with like the human in the loop. I think the way I framed it was
Ben Callahan (28:31)
list them out in a nice way. That used to take me like half a day or a whole day to like put those types of things together that I can now automate and say, hey, look at all the changes over the last week and like give me a nice executive summary for my boss who wants like three bullet points that are one sentence each. That used to take me so long. And I can automate those things now that allow me to work on the things that are important. Because I really agree with like the human in the loop. I think the way I framed it was
ep073 speaker raw (28:31)
list them out in a nice way. That used to take me like half a day or a whole day to like put those types of things together that I can now automate and say, hey, look at all the changes over the last week and like give me a nice executive summary for my boss who wants like three bullet points that are one sentence each. That used to take me so long. And I can automate those things now that allow me to work on the things that are important. Because I really agree with like the human in the loop. I think the way I framed it was
ep073 screen raw (28:59)
A lot of this stuff needs like a human touch. you know, I don't want to automate, adopt or feedback and have my users talking to an AI and then distilling that information and giving it to me. Like I want to talk to them. I want to automate stuff that allows me to have time to go spend an hour sitting with my users and understanding the problems that they're having. So that's kind of the way I see it. I think these things will live in conjunction together.
Ben Callahan (28:59)
A lot of this stuff needs like a human touch. you know, I don't want to automate, adopt or feedback and have my users talking to an AI and then distilling that information and giving it to me. Like I want to talk to them. I want to automate stuff that allows me to have time to go spend an hour sitting with my users and understanding the problems that they're having. So that's kind of the way I see it. I think these things will live in conjunction together.
ep073 speaker raw (28:59)
A lot of this stuff needs like a human touch. you know, I don't want to automate, adopt or feedback and have my users talking to an AI and then distilling that information and giving it to me. Like I want to talk to them. I want to automate stuff that allows me to have time to go spend an hour sitting with my users and understanding the problems that they're having. So that's kind of the way I see it. I think these things will live in conjunction together.
ep073 screen raw (29:28)
You know, use AI to automate things, to automate things that might be hard to automate that you just never had time, right? It's like, I don't have time to write these scripts and do these things. Maybe I can use AI to like help me write that script. ⁓ So yeah, that's, yeah. But I think I mostly agree with, I think everybody's almost on the same page it sounds, which is nice to hear. That reminds me of some of the conversations we had last session with Doug, who's here actually too, co-hosting around.
Ben Callahan (29:28)
You know, use AI to automate things, to automate things that might be hard to automate that you just never had time, right? It's like, I don't have time to write these scripts and do these things. Maybe I can use AI to like help me write that script. ⁓ So yeah, that's, yeah. But I think I mostly agree with, I think everybody's almost on the same page it sounds, which is nice to hear. That reminds me of some of the conversations we had last session with Doug, who's here actually too, co-hosting around.
ep073 speaker raw (29:28)
You know, use AI to automate things, to automate things that might be hard to automate that you just never had time, right? It's like, I don't have time to write these scripts and do these things. Maybe I can use AI to like help me write that script. ⁓ So yeah, that's, yeah. But I think I mostly agree with, I think everybody's almost on the same page it sounds, which is nice to hear. That reminds me of some of the conversations we had last session with Doug, who's here actually too, co-hosting around.
ep073 screen raw (29:57)
extreme customer support in systems. like you want to have that conversation. It's okay for them to reach out to you because you're going to learn so much in that moment in that conversation. Davy, anything to chime in here?
Ben Callahan (29:57)
extreme customer support in systems. like you want to have that conversation. It's okay for them to reach out to you because you're going to learn so much in that moment in that conversation. Davy, anything to chime in here?
ep073 speaker raw (29:57)
extreme customer support in systems. like you want to have that conversation. It's okay for them to reach out to you because you're going to learn so much in that moment in that conversation. Davy, anything to chime in here?
ep073 screen raw (30:12)
I was thinking like that I read it. read a tweet that said like this is it's it's amazing that Kron is having like it's Renaissance moment and like 2026. ⁓ yes, like that's where my mind goes back to. Gosh. Yeah. Lauren, what's going on?
Ben Callahan (30:12)
I was thinking like that I read it. read a tweet that said like this is it's it's amazing that Kron is having like it's Renaissance moment and like 2026. ⁓ yes, like that's where my mind goes back to. Gosh. Yeah. Lauren, what's going on?
ep073 speaker raw (30:12)
I was thinking like that I read it. read a tweet that said like this is it's it's amazing that Kron is having like it's Renaissance moment and like 2026. ⁓ yes, like that's where my mind goes back to. Gosh. Yeah. Lauren, what's going on?
ep073 screen raw (30:34)
Hi, yeah, thanks. Yeah, super insightful. And I think like the obvious thing, it seems to be kind of a shared consensus about human relationships, like the loss of connection, what AI can kind of get in the way of us communicating and kind of getting on the same page and that knowledge transfer. But I think that we've been using it to train our team, because I think, know, new tools come out all the time.
ep073 speaker raw (30:34)
Hi, ⁓ yeah, thanks. ⁓ Yeah, super insightful. And I think like the obvious thing, it seems to be kind of a shared consensus about human relationships, like the loss of connection, what AI can ⁓ kind of get in the way of us communicating and kind of getting on the same page and that knowledge transfer. But I think that ⁓ we've been using it to train our team, because I think, know, new tools come out all the time.
Ben Callahan (30:34)
Hi, ⁓ yeah, thanks. ⁓ Yeah, super insightful. And I think like the obvious thing, it seems to be kind of a shared consensus about human relationships, like the loss of connection, what AI can ⁓ kind of get in the way of us communicating and kind of getting on the same page and that knowledge transfer. But I think that ⁓ we've been using it to train our team, because I think, know, new tools come out all the time.
ep073 screen raw (31:04)
So we've been working on like a Wiki, whenever someone learns about a different, ⁓ you know, feature or a new skill, it's actually kind of brought our team quite closer because it's meant that ⁓ we have this kind of shared knowledge. So I think the human relationships is, ⁓ yeah, it can kind of swing both ways. Just wanted to add that. Can you say a little bit more? Like, do you have an example or something?
ep073 speaker raw (31:04)
So we've been working on like a Wiki, whenever someone learns about a different, ⁓ you know, feature or a new skill, it's actually kind of brought our team quite closer because it's meant that we have this kind of shared knowledge. So I think the human relationships is, ⁓ yeah, it can kind of swing both ways. Just wanted to add that. Can you say a little bit more? Like, do you have an example or something?
Ben Callahan (31:04)
So we've been working on like a Wiki, whenever someone learns about a different, ⁓ you know, feature or a new skill, it's actually kind of brought our team quite closer because it's meant that ⁓ we have this kind of shared knowledge. So I think the human relationships is, ⁓ yeah, it can kind of swing both ways. Just wanted to add that. Can you say a little bit more? Like, do you have an example or something?
ep073 screen raw (31:32)
that you can think of specifically where you feel like that sort of experimentation or use of AI forced your team to kind of reconnect in a new way or? So ⁓ because there's so much to learn and it's constantly changing, I've essentially built out like a dashboard and all of the team have profiles on this, on this like Wiki. And so there's kind of progress reports. They can see how much content they're getting through.
Ben Callahan (31:32)
that you can think of specifically where you feel like that sort of experimentation or use of AI forced your team to kind of reconnect in a new way or? So ⁓ because there's so much to learn and it's constantly changing, I've essentially built out like a dashboard and all of the team have profiles on this, on this like Wiki. And so there's kind of progress reports. They can see how much content they're getting through.
ep073 speaker raw (31:32)
that you can think of specifically where you feel like that sort of experimentation or use of AI forced your team to kind of reconnect in a new way or? So ⁓ because there's so much to learn and it's constantly changing, I've essentially built out like a dashboard and all of the team have profiles on this, on this like Wiki. And so there's kind of progress reports. They can see how much content they're getting through.
ep073 screen raw (32:02)
So for example, we've been exploring how to build out our token architecture and we've done all these workshops with like our devs and trying to understand, okay, what's working, what isn't working. We really want to be able to bake in kind of more of a multi-tier ⁓ system. And so through those workshops, we've then captured those meetings in this Wiki so that we've got that information. And then we've got this kind of source of truth just as a page on the Wiki. that...
Ben Callahan (32:02)
So for example, we've been exploring how to build out our token architecture and we've done all these workshops with like our devs and try and understand, okay, what's working, what isn't working. We really wanna be able to bake in kind of more of a multi-tier ⁓ system. And so through those workshops, we've then captured those meetings in this Wiki so that we've got that information. And then we've got this kind of source of truth just as a page on the Wiki. So that...
ep073 speaker raw (32:02)
So for example, we've been exploring how to build out our token architecture. And we've done all these workshops with like our devs and try and understand, okay, what's working, what isn't working. We really want to be able to bake in kind of more of a multi-tier ⁓ system. And so through those workshops, we've then captured those meetings in this Wiki so that we've got that information. And then we've got this kind of source of truth just as a page on the Wiki. that...
ep073 screen raw (32:28)
all of the team can now look through it. And if they haven't read it, then that shows up on their progress and says, you haven't kind of checked in on this information. So it's kind of just assisting us with that knowledge transfer ⁓ without kind of beating us with a stick. Gotcha. Nice. Thank you. Kalista, welcome. Hi. Hi, thank you. I wanted to tie back to something Sarah and both Randy were kind of
Ben Callahan (32:28)
all of the team can now look through it. And if they haven't read it, then that shows up on their progress and says, you haven't kind of checked in on this information. So it's kind of just assisting us with that knowledge transfer ⁓ without kind of beating us with a stick. Gotcha. Nice. Thank you. Kalista, welcome. Hi. Hi, thank you. I wanted to tie back to something Sarah and both Randy were kind of
ep073 speaker raw (32:28)
all of the team can now look through it. And if they haven't read it, then that shows up on their progress and says, you haven't kind of checked in on this information. So it's kind of just assisting us with that knowledge transfer ⁓ without kind of beating us with a stick. Gotcha. Nice. Thank you. Kalista, welcome. Hi. Hi, thank you. I wanted to tie back to something Sarah and both Randy were kind of
ep073 screen raw (32:56)
had the similar sentiment and that automation is maybe the final part of all of these processes. And I went to a panel yesterday where Mark Siafrani from Kin Insurance made the analogy that right now the way that maybe designers should be looking at AI in general is he used the analogy of the Beatles with music, which really stuck with me of how it was a lot of experimentation and play.
Ben Callahan (32:56)
had the similar sentiment and that automation is maybe the final part of all of these processes. And I went to a panel yesterday where Mark Siefrani from Kin Insurance made the analogy that right now the way that maybe designers should be looking at AI in general is he used the analogy of the Beatles with music, which really stuck with me of how it was a lot of experimentation and play.
ep073 speaker raw (32:56)
had the similar sentiment and that automation is maybe the final part of all of these processes. And I went to a panel yesterday where Mark Siefrani from Kin Insurance made the analogy that right now the way that maybe designers should be looking at AI in general is he used the analogy of the Beatles with music, which really stuck with me of how it was a lot of experimentation and play.
ep073 screen raw (33:25)
And I think that it would be really interesting to see a lot of these organizations that are pushing AI into design teams, allow teams to have structured playtime with just exploring different MD file formats and different ways to prompt or build prompt templates across their team. Because right now, of course, we don't have those industry standards for any of this.
Ben Callahan (33:25)
And I think that it would be really interesting to see a lot of these organizations that are pushing AI into design teams, allow teams to have structured playtime with just exploring different MD file formats and different ways to prompt or build prompt templates across their team. Because right now, of course, we don't have those industry standards for any of this.
ep073 speaker raw (33:25)
And I think that it would be really interesting to see a lot of these organizations that are pushing AI into design teams, allow teams to have structured playtime with just exploring different MD file formats and different ways to prompt or build prompt templates across their team. Because right now, of course, we don't have those industry standards for any of this.
ep073 screen raw (33:53)
because we're all kind of at the starting line together. And I think that organizations that we see do push forward on that structured play time are going to be the ones that have the ability to really shape the industry standards of those guardrails. Whereas the other organizations are going to be learning from the ones that have already spent time playing and experimenting there. I love that. Yeah. And I saw some folks talking about ⁓ hackathons, Elizabeth. Yes. Super cool.
Ben Callahan (33:53)
because we're all kind of at the starting line together. And I think that organizations that we see do push forward on that structured play time are going to be the ones that have the ability to really shape the industry standards of those guardrails. Whereas the other organizations are going to be learning from the ones that have already spent time playing and experimenting there. I love that. Yeah. And I saw some folks talking about ⁓ hackathons, Elizabeth. Yes. Super cool.
ep073 speaker raw (33:53)
because we're all kind of at the starting line together. And I think that organizations that we see do push forward on that structured play time are going to be the ones that have the ability to really shape the industry standards of those guardrails. Whereas the other organizations are going to be learning from the ones that have already spent time playing and experimenting there. I love that. Yeah. And I saw some folks talking about ⁓ hackathons, Elizabeth. Yes. Super cool.
ep073 screen raw (34:23)
Thank you, Kalissa, that's a great insight. ⁓ While we're getting to some of these other raised hands, there's a question down here which says, what are you already automating with AI? So if you have a moment, jump into the FigJam, you can drag out a sticky note from this one and just drop in. I wanna hear the things that you're using AI to already automate. So do that in the FigJam and then we'll come back to that.
Ben Callahan (34:23)
Thank you, Kalissa, that's a great insight. ⁓ While we're getting to some of these other raised hands, there's a question down here which says, what are you already automating with AI? So if you have a moment, jump into the FigJam, you can drag out a sticky note from this one and just drop in. I wanna hear the things that you're using AI to already automate. So ⁓ do that in the FigJam and then we'll come back to that.
ep073 speaker raw (34:23)
Thank you, Kalissa, that's a great insight. ⁓ While we're getting to some of these other raised hands, there's a question down here which says, what are you already automating with AI? So if you have a moment, jump into the FigJam, you can drag out a sticky note from this one and just drop in. I wanna hear the things that you're using AI to already automate. So ⁓ do that in the FigJam and then we'll come back to that.
ep073 screen raw (34:48)
Guy, jump joined halfway through,
Ben Callahan (34:48)
Guy, jump joined halfway through,
ep073 speaker raw (34:48)
Guy, jump joined halfway through,
ep073 screen raw (34:51)
please tell me if I'm already repeating things that people talked about. But I wanted to react to something that Kalista just said about the, and other people said in the chat as well about the experimentation. And this is exactly where I feel my organization was in last year. Like we were really told to go try every tool, experiment, find things that work for you, find the sort of the edges of where it makes sense. This year, it's a complete change. Like there's still,
Ben Callahan (34:51)
please tell me if I'm already repeating things that people talked about. But I wanted to react to something that Kalista just said about the, and other people said in the chat as well about the experimentation. And this is exactly where I feel my organization was in last year. Like we were really told to go try every tool, experiment, find things that work for you, find the sort of the edges of where it makes sense. This year, it's a complete change. Like there's still,
ep073 speaker raw (34:51)
please tell me if I'm already repeating things that people talked about. But I wanted to react to something that Kalista just said about the, and other people said in the chat as well about the experimentation. And this is exactly where I feel my organization was in last year. Like we were really told to go try every tool, experiment, find things that work for you, find the sort of the edges of where it makes sense. This year, it's a complete change. Like there's still,
ep073 screen raw (35:22)
hey, try this new thing, but there's a lot of ⁓ leadership has set the expectations or they have set the, said, ⁓ use these tools and get us 20 % efficiency. Like all those things, there's already been a decision of how well they're gonna be working. ⁓ Claw design was announced Friday. ⁓
Ben Callahan (35:22)
hey, try this new thing, but there's a lot of ⁓ leadership has set the expectations or they have set the, said, ⁓ use these tools and get us 20 % efficiency. Like all those things, there's already been a decision of how well they're gonna be working. Claw design was announced Friday. ⁓
ep073 speaker raw (35:22)
hey, try this new thing, but there's a lot of ⁓ leadership has set the expectations or they have set the, said, ⁓ use these tools and get us 20 % efficiency. Like all those things, there's already been a decision of how well they're gonna be working. Claw design was announced Friday. ⁓
ep073 screen raw (35:51)
We already have people who say, we can use it to tie it to cloud code and do all those things. And we haven't even tried it. I'm curious to see how that is affecting people's work. The fact that they have to meet this yet another imaginary number that someone defined on this thing. What, Guy, what do you think they looked at?
Ben Callahan (35:51)
We already have people who say, we can use it to tie it to cloud code and do all those things. And we haven't even tried it. I'm curious to see how that is affecting people's work. The fact that they have to meet this yet another imaginary number that someone defined on this thing. What, Guy, what do you think they looked at?
ep073 speaker raw (35:51)
We already have people who say, we can use it to tie it to cloud code and do all those things. And we haven't even tried it. I'm curious to see how that is affecting people's work. The fact that they have to meet this yet another imaginary number that someone defined on this thing. What, Guy, what do you think they looked at?
ep073 screen raw (36:15)
the experimentation from last year and said, okay, that's enough. It's time to put it into practice. what? No, no, that would be too smart. ⁓ I think it's another symptom of the ⁓ CEO said something. it. Okay. Yeah. Anytime we bring AI into the conversation on the question, we get these, you know, examples of executives, you know, seeing a demo, reading a post.
Ben Callahan (36:15)
the experimentation from last year and said, okay, that's enough. It's time to put it into practice. what? No, no, that would be too smart. ⁓ I think it's another symptom of the ⁓ CEO said something. Got it. Okay. Yeah. Anytime we bring AI into the conversation on the question, we get these, you know, examples of executives, you know, seeing a demo, reading a post.
ep073 speaker raw (36:15)
the experimentation from last year and said, okay, that's enough. It's time to put it into practice. what? No, no, that would be too smart. ⁓ I think it's another symptom of the ⁓ CEO said something. Got it. Okay. Yeah. Anytime we bring AI into the conversation on the question, we get these, you know, examples of executives, you know, seeing a demo, reading a post.
Ben Callahan (36:43)
and just asking for that to be the reality immediately. So yeah, that's a common one. Donnie, did we get it working? Let's see. Did I get it working? Yes. Oh my god. done. Ben knows how much I hate audio. We together to try to fix my podcast downstairs. But the question I had actually was about, Davy mentioned earlier, and we're talking about skills in the chat as well.
ep073 screen raw (36:43)
and just asking for that to be the reality immediately. So yeah, that's a common one. Donnie, did we get it working? Let's see. Did I get it working? Yes. my god. done. Ben knows how much I hate audio. We together to try to fix my podcast downstairs. But ⁓ the question I had actually was about, Davy mentioned earlier, and we're talking about skills in the chat as well. ⁓
ep073 speaker raw (36:43)
and just asking for that to be the reality immediately. So yeah, that's a common one. Donnie, did we get it working? Let's see. Did I get it working? Yes. my god. done. Ben knows how much I hate audio. We together to try to fix my podcast downstairs. But ⁓ the question I had actually was about, Davy mentioned earlier, and we're talking about skills in the chat as well. ⁓
ep073 screen raw (37:12)
I was at a meetup last night where we're talking about skills versus MCP. And the question is, from my perspective, is if we are focusing on the skills as a direction to take for people to activate something, my understanding was that the skills are effectively impossible to update because they're all local to people's machines. Where I would have found an MCP to be more of a thing that could update
Ben Callahan (37:12)
I was at a meetup last night where we're talking about skills versus MCP. And the question is, from my perspective, is if we are focusing on the skills as a direction to take for people to activate something, my understanding was that the skills are effectively impossible to update because they're all local to people's machines. Where I would have found an MCP to be more of a thing that could update
ep073 speaker raw (37:12)
I was at a meetup last night where we're talking about skills versus MCP. And the question is, from my perspective, is if we are focusing on the skills as a direction to take for people to activate something, my understanding was that the skills are effectively impossible to update because they're all local to people's machines. Where I would have found an MCP to be more of a thing that could update
ep073 screen raw (37:41)
on the fly and be more remote. like when that needs to, you when we have updates to the system, guidelines and things like that, connecting to the MCP from a source of truth, like we like to call it, I think would make more sense than the skills. So if for the folks that are using skills as their main mechanism to provide guidelines, how are you keeping those things ⁓ updated in sync across organizations? Great question. David, you want to jump into that?
Ben Callahan (37:42)
on the fly and be more remote. like when that needs to, you when we have updates to the system, guidelines and things like that, connecting to the MCP from a source of truth, like we like to call it, I think would make more sense than the skills. So if for the folks that are using skills as their main mechanism to provide guidelines, how are you keeping those things ⁓ updated in sync across organizations? Great question. David, you want to jump into that?
ep073 speaker raw (37:42)
on the fly and be more remote. like when that needs to, you when we have updates to the system, guidelines and things like that, connecting to the MCP from a source of truth, like we like to call it, I think would make more sense than the skills. So if for the folks that are using skills as their main mechanism to provide guidelines, how are you keeping those things updated in sync across organizations? Great question. David, you want to jump into that?
ep073 screen raw (38:10)
Yeah, I think you could have them on a local level or a project level. But yeah, we've been putting them on the repo levels in different directories. And that's worked out pretty good. But we're still trying to figure out what is the best way to distribute them. I don't believe there is awareness in.
Ben Callahan (38:10)
Yeah, I think you could have them on a local level or a project level. But yeah, we've been putting them on the repo levels in different directories. And that's worked out pretty good. But we're still trying to figure out what is the best way to distribute them. I don't believe there is awareness in
ep073 speaker raw (38:10)
Yeah, I think you could have them on a local level or a project level. But yeah, we've been putting them on the repo levels in different directories. And that's worked out pretty good. But we're still trying to figure out what is the best way to distribute them. I don't believe there is awareness in.
ep073 screen raw (38:34)
Figma, for instance, like if I publish an update to a skill like you downstream and your CLI will not know that I published this. So it's sort of like a, we need to figure out how that that works. And, ⁓ also like in the prototyping world, for those of us that are prototyping in tools like V zero or Replet, that is also like a problem right now where we're trying to figure out, ⁓ how can we distribute skills across like a workspace versus.
Ben Callahan (38:34)
Figma, for instance, like if I publish an update to a skill like you downstream and your CLI will not know that I published this. So it's sort of like a, we need to figure out how that that works. And, ⁓ also like in the prototyping world, for those of us that are prototyping in tools like V zero or Replet, that is also like a problem right now where we're trying to figure out, ⁓ how can we distribute skills across like a workspace versus.
ep073 speaker raw (38:34)
Figma, for instance, like if I publish an update to a skill like you downstream and your CLI will not know that I published this. So it's sort of like a, we need to figure out how that that works. And, ⁓ also like in the prototyping world, for those of us that are prototyping in tools like V zero or Replet, that is also like a problem right now where we're trying to figure out, ⁓ how can we distribute skills across like a workspace versus.
ep073 screen raw (39:03)
Hey, Donnie, I've updated my skill. Can you go pull this down? Well, it's not the same problem that we have right now about having some new version of a thing that comes out. And then you got to tell the world that there's a new version. And some people listen to it. Some people don't listen to it. Some people do the command. Some people don't. So that's the thing that I feel like the MCP kind of solves for, where connect to it once.
Ben Callahan (39:03)
Hey, Donnie, I've updated my skill. Can you go pull this down? Well, it's not the same problem that we have right now about having some new version of a thing that comes out. And then you got to tell the world that there's a new version. And some people listen to it. Some people don't listen to it. Some people do the command. Some people don't. So that's the thing that I feel like the MCP kind of solves for, where connect to it once.
ep073 speaker raw (39:03)
Hey, Donnie, I've updated my skill. Can you go pull this down? Well, it's not the same problem that we have right now about having some new version of a thing that comes out. And then you got to tell the world that there's a new version. And some people listen to it. Some people don't listen to it. Some people do the command. Some people don't. So that's the thing that I feel like the MCP kind of solves for, where connect to it once.
ep073 screen raw (39:30)
And then you always get the latest whatever in terms of guidance and stuff like that. So that's why I keep seeing the skills stuff. I'm like, I get the skills being good for your local, what I'm working on in my space, but for the organizational one, I have yet to seen a good, I guess, ecosystem that supports updating of skills in a way that people don't need to think about it so much, but you do get the latest and greatest. You know what mean? Yeah.
Ben Callahan (39:30)
And then you always get the latest whatever in terms of guidance and stuff like that. So that's why I keep seeing the skills stuff. I'm like, I get the skills being good for your local, what I'm working on in my space, but for the organizational one, I have yet to seen a good, I guess, ecosystem that supports updating of skills in a way that people don't need to think about it so much, but you do get the latest and greatest. You know what mean? Yeah.
ep073 speaker raw (39:30)
And then you always get the latest whatever in terms of guidance and stuff like that. So that's why I keep seeing the skills stuff. I'm like, I get the skills being good for your local, what I'm working on in my space, but for the organizational one, I have yet to seen a good, I guess, ecosystem that supports updating of skills in a way that people don't need to think about it so much, but you do get the latest and greatest. You know what mean? Yeah.
ep073 screen raw (40:00)
Peter, did you have something?
Ben Callahan (40:00)
Peter, did you have something?
ep073 speaker raw (40:00)
Peter, did you have something?
Ben Callahan (40:01)
Yeah, I mean, was going to try to take my best stab at Dottie's kind of question. now, granted, I'm using a different tool with using Kero versus Cloud Code. So skills slightly different than some of the things, or at least the naming is slightly different. So this could be off. But the key for me that I've been able to get is you use both. So I use the MCP as the kind of domain knowledge.
ep073 speaker raw (40:01)
Yeah, I mean, was going to try to take my best stab at Dottie's kind of question. now, granted, I'm using a different tool with using Kero versus Cloud Code. So skills slightly different than some of the things, or at least the naming is slightly different. So this could be off. But the key for me that I've been able to get is you use both. So I use the MCP as the kind of domain knowledge.
ep073 screen raw (40:02)
Yeah, I mean, was going to try to take my best stab at Donnie's kind of question. now granted, I'm using a different tool with using Kero versus Cloud Code. So skills slightly different than some of the things, or at least the naming is slightly different. So this could be off. But the key for me that I've been able to get is you use both. So I use the MCP as the kind of domain knowledge.
container, is basically the library of all the information. And then all of my custom agents basically plug into that MCP to pull their skills and all those kinds of things. So really kind of the MCP acts as a library. So every agent has access to the information on that MCP, but certain agents have specific, basically are inherently tuned in.
Ben Callahan (40:30)
container, is basically the library of all the information. And then all of my custom agents basically plug into that MCP to pull their skills and all those kinds of things. So really kind of the MCP acts as a library. So every agent has access to the information on that MCP, but certain agents have specific, basically are inherently tuned in.
ep073 speaker raw (40:30)
container, is basically the library of all the information. And then all of my custom agents basically plug into that MCP to pull their skills and all those kinds of things. So really kind of the MCP acts as a library. So every agent has access to the information on that MCP, but certain agents have specific, basically are inherently tuned in.
ep073 screen raw (40:55)
to specific aspects of that MCP, specific aspects of the documentation that's stored on those MCPs. So, and I think this has been echoed in a lot of folks that I follow is that you don't have like the MCP or the agent, but you use both in parallel, where you use kind of the MCP as kind of like your hub, but your agent has specific aspects of it to kind of focus on. Yeah. Awesome. Thank you, Peter.
Ben Callahan (40:55)
to specific aspects of that MCP, specific aspects of the documentation that's stored on those MCPs. So, and I think this has been echoed in a lot of folks that I follow is that you don't have like the MCP or the agent, but you use both in parallel, where you use kind of the MCP as kind of like your hub, but your agent has specific aspects of it to kind of focus on. Yeah. Awesome. Thank you, Peter.
ep073 speaker raw (40:55)
to specific aspects of that MCP, specific aspects of the documentation that's stored on those MCPs. So, and I think this has been echoed in a lot of folks that I follow is that you don't have like the MCP or the agent, but you use both in parallel, where you use kind of the MCP as kind of like your hub, but your agent has specific aspects of it to kind of focus on. Yeah. Awesome. Thank you, Peter.
ep073 screen raw (41:23)
If really, really quick, just, just to chime in on that too, I actually just did this today. We've been working on it the last couple of weeks. So we're using cursor and Claude in our org. Both of those tools have marketplaces and plugins that you can add at either an enterprise level or public rate. Like Figma has a plugin, et cetera, et cetera. We basically just hosted a design system plugin in our internal marketplace for cursor and Claude.
Ben Callahan (41:23)
If really, really quick, just, I just to chime in on that too. I actually just did this today. We've been working on it the last couple of weeks. So we're using cursor and Claude in our org. Both of those tools have marketplaces and plugins that you can add at either an enterprise level or public rate. Like Figma has a plugin, et cetera, et cetera. We basically just hosted a design system plugin in our internal marketplace for cursor and Claude.
ep073 speaker raw (41:23)
If really, really quick, just, I just to chime in on that too. I actually just did this today. We've been working on it the last couple of weeks. So we're using cursor and Claude in our org. Both of those tools have marketplaces and plugins that you can add at either an enterprise level or public rate. Like Figma has a plugin, et cetera, et cetera. We basically just hosted a design system plugin in our internal marketplace for cursor and Claude.
ep073 screen raw (41:48)
And as I update skills and MCPs and agents and all this stuff in the plugin, that just gets distributed out to everybody in our enterprise that's on our account. So like they just get the most up-to-date versions of those skills. And we can actually enable them and put them into people's workspaces automatically if we really wanted to. We're not doing that, but that's also a way that if you are in an org that's using those tools. ⁓
Ben Callahan (41:48)
And as I update skills and MCPs and agents and all this stuff in the plugin, that just gets distributed out to everybody in our enterprise that's on our account. So like they just get the most up-to-date versions of those skills. And we can actually enable them and put them into people's workspaces automatically if we really wanted to. We're not doing that, but that's also a way that if you are in an org that's using those tools. ⁓
ep073 speaker raw (41:48)
And as I update skills and MCPs and agents and all this stuff in the plugin, that just gets distributed out to everybody in our enterprise that's on our account. So like they just get the most up-to-date versions of those skills. And we can actually enable them and put them into people's workspaces automatically if we really wanted to. We're not doing that, but that's also a way that if you are in an org that's using those tools. ⁓
ep073 screen raw (42:15)
I would highly suggest looking into plugins ⁓ for those, you'll find a lot of good information. Thanks, Stephen. Randy, what's up? Hey, so one of the things that I really like AI for, like pushing towards automation, is actually using AI to create tools to help with other tasks that I need to do. Like one of the reasons why I kind of got into doing websites in the first place is because I'm like a builder and a maker.
Ben Callahan (42:15)
I would highly suggest looking into plugins ⁓ for those, you'll find a lot of good information. Thanks, Stephen. Randy, what's up? Hey, so one of the things that I really like AI for, like pushing towards automation, is actually using AI to create tools to help with other tasks that I need to do. Like one of the reasons why I kind of got into doing websites in the first place is because I'm like a builder and a maker.
ep073 speaker raw (42:15)
I would highly suggest looking into plugins ⁓ for those, you'll find a lot of good information. Thanks, Stephen. Randy, what's up? Hey, so one of the things that I really like AI for, like pushing towards automation, is actually using AI to create tools to help with other tasks that I need to do. Like one of the reasons why I kind of got into doing websites in the first place is because I'm like a builder and a maker.
ep073 screen raw (42:44)
And I think that like, need to remember that like, we can create these intermediary steps to help with automation, to help with very frequent tasks. And so that I think is one of the most exciting parts of AI because like I had AI help me build like a type scale tool, helped me build a tool for doing like fluid type and spacing because I don't want to do that math. Like I went to art school. I don't want to do that.
Ben Callahan (42:44)
And I think that like, need to remember that like, we can create these intermediary steps to help with automation, to help with very frequent tasks. And so that I think is one of the most exciting parts of AI because like I had AI help me build like a type scale tool, helped me build a tool for doing like fluid type and spacing because I don't want to do that math. Like I went to art school. I don't want to do that.
ep073 speaker raw (42:44)
And I think that like, we need to remember that like, can create these intermediary steps to help with automation, to help with very frequent tasks. ⁓ And so that I think is one of the most exciting parts of AI, because like I had AI help me build like a type scale tool helped me build a tool for doing like fluid type and spacing, because I want to do that math. Like I went to art school. I don't want to do that.
ep073 screen raw (43:11)
Yeah, I love that. Randy over in Redwoods, we've had a handful of folks build many little sort of custom plugins in Figma just to help them solve this very specific problem. They're not developers, they're designers, but they're building these things with the use of AI. And Davy, I think that's something that you talked about too when we first chatted about this is like these very tiny sort of like tools that you can just on the fly come up with very quickly. Yeah, that's right. Like these tools or any sort of like
Ben Callahan (43:11)
Yeah, I love that. Randy over in Redwoods, we've had a handful of folks build many little sort of custom plugins in Figma just to help them solve this very specific problem. They're not developers, they're designers, but they're building these things with the use of AI. And Davy, I think that's something that you talked about too when we first chatted about this is like these very tiny sort of like tools that you can just on the fly come up with very quickly. Yeah, that's right. Like these tools or any sort of like
ep073 speaker raw (43:11)
Yeah, I love that. Randy over in Redwoods, we've had a handful of folks build many little sort of custom plugins in Figma just to help them solve this very specific problem. They're not developers, they're designers, but they're building these things with the use of AI. And Davy, I think that's something that you talked about too when we first chatted about this is like these very tiny sort of like tools that you can just on the fly come up with very quickly. Yeah, that's right. Like these tools or any sort of like
ep073 screen raw (43:40)
Simple like empty scripts that you could run and eventually hopefully they could be useful to somebody else, but they're always ⁓ more so created so I could do something again very, very quickly. Yeah, yeah guy. I went through. React to something that Keisha put in the chat about like being pulled in 100 different directions. Every day there's a new tool coming out. ⁓ We're feeling it very, very much.
Ben Callahan (43:40)
simple like MPM scripts that you could run and eventually hopefully they could be useful to somebody else, but they're always ⁓ more so created so I could do something again very, very quickly. Yeah, yeah. Guy. I went through. React to something that Keisha put in the chat about like being pulled in 100 different directions with every day there's a new tool coming out. ⁓ We're feeling it very, very much.
ep073 speaker raw (43:40)
simple like MTM scripts that you could run and eventually hopefully they could be useful to somebody else, but they're always ⁓ more so created so I could do something again very, very quickly. Yeah, yeah. Guy. I went through. React to something that Keisha put in the chat about like being pulled in 100 different directions with every day there's a new tool coming out. ⁓ We're feeling it very, very much.
ep073 screen raw (44:09)
on our team with our organization is using v0, Claude code, cursor, lovable, ⁓ Figma make, and now with this new Claude design coming in, shameless plug for my article that just came out two days ago about the hidden cost of saying yes to things. These are things that we haven't even said yes to. The organization told us that we need to support.
Ben Callahan (44:09)
on our team with our organization is using v0, Claude code, cursor, lovable, Figma make, and now with this new Claude design coming in, ⁓ shameless plug for my article that just came out two days ago about the hidden cost of saying yes to things. These are things that we haven't even said yes to. The organization told us that we need to support.
ep073 speaker raw (44:09)
on our team with our organization is using v0, Claude code, cursor, lovable, ⁓ Figma make, and now with this new Claude design coming in, shameless plug for my article that just came out two days ago about the cost of saying yes to things. These are things that we haven't even said yes to. The organization told us that we need to support.
ep073 screen raw (44:37)
from a design system perspective, or if we don't support, then someone else is gonna be using their own solution with those tools. ⁓ And we're really feeling now the stress of, ⁓ the expectation of product design teams is that we're the ones who are supporting V0's horrible support for ingesting a design system.
Ben Callahan (44:37)
from a design system perspective, or if we don't support, then someone else is gonna be using their own solution with those tools. ⁓ And we're really feeling now the stress of, ⁓ the expectation of product design teams is that we're the ones who are supporting V0's horrible support for ingesting a design system.
ep073 speaker raw (44:37)
from a design system perspective, or if we don't support, then someone else is gonna be using their own solution with those tools. ⁓ And we're really feeling now the stress of, ⁓ the expectation of product design teams is that we're the ones who are supporting V0's horrible support for ingesting a design system.
ep073 screen raw (45:02)
And we're the ones who have to go and talk to them and maintain that relationship and maintain that level of quality. And it's something that we would never sign up for if we had the choice. Yeah. Thanks, Guy. I think that that thread there, Keisha, appreciate you sharing that. I mean, I think in my mind, the way we address this challenge of feeling completely overwhelmed is by having community like this. Right. Because no one person can possibly stay up to date on all this stuff. So if we can bring our
Ben Callahan (45:02)
And we're the ones who have to go and talk to them and maintain that relationship and maintain that level of quality. And it's something that we would never sign up for if we had the choice. Yeah. Thanks, Guy. I think that that thread there, Keisha, appreciate you sharing that. I mean, I think in my mind, the way we address this challenge of feeling completely overwhelmed is by having community like this. Right. Because no one person can possibly stay up to date on all this stuff. So if we can bring our
ep073 speaker raw (45:02)
And we're the ones who have to go and talk to them and maintain that relationship and maintain that level of quality. And it's something that we would never sign up for if we had the choice. Yeah. Thanks, Guy. I think that that thread there, Keisha, appreciate you sharing that. I mean, I think in my mind, the way we address this challenge of feeling completely overwhelmed is by having community like this. Right. Because no one person can possibly stay up to date on all this stuff. So if we can bring our
Ben Callahan (45:31)
minds together in these kinds of conversations. That's how we do it. Doug, what's up? I have a very specific question. And if it's too specific, it can totally be tabled. But Dave, you mentioned that you're using an MCP for your design documentation. I think that was you that left the comment. Yeah. I'm trying to, like, I cannot find the right search term. Everything I'm coming up with is talking about how to use Figma MCP or how to use it. I just want to, like, what are the best tools to expose your design documentation?
ep073 screen raw (45:31)
minds together in these kinds of conversations. That's how we do it. Doug, what's up? I have a very specific question. And if it's too specific, it can totally be tabled. But Dave, you mentioned that you're using an MCP for your design documentation. ⁓ I think that was you that left the comment. Yeah. I'm trying to, like, I cannot find the right search term. Everything I'm coming up with is talking about how to use Figma MCP or how to use it. I just want to, like, what are the best tools to expose your design documentation?
ep073 speaker raw (45:31)
minds together in these kinds of conversations. That's how we do it. Doug, what's up? I have a very specific question. And if it's too specific, it can totally be tabled. But Dave, you mentioned that you're using an MCP for your design documentation. ⁓ I think that was you that left the comment. Yeah. I'm trying to, like, I cannot find the right search term. Everything I'm coming up with is talking about how to use Figma MCP or how to use it. I just want to, like, what are the best tools to expose your design documentation?
Ben Callahan (46:01)
to an AI in an MCP server? Do you have just a couple tool calls, then it does everything? Is there any guidance on that? Or even an article I can read? Yeah, I'll have to try to find something after I give my response. But I think a lot of it is with ⁓ setting up your Doc site, for instance, for structured data. So we did this huge vision quest late last year to figure out. ⁓
ep073 screen raw (46:01)
to an AI in an MCP server? Do you have just a couple tool calls, then it does everything? Is there any guidance on that? Or even an article I can read? Yeah, I'll have to try to find something after I give my response. But I think a lot of it is with setting up your Doc site, for instance, for structured data. So we did this huge vision quest late last year to figure out. ⁓
ep073 speaker raw (46:01)
to an AI in an MCP server? Do you have just a couple tool calls, then it does everything? Is there any guidance on that? Or even an article I can read? Yeah, I'll have to try to find something after I give my response. But I think a lot of it is with ⁓ setting up your Doc site, for instance, for structured data. So we did this huge vision quest late last year to figure out. ⁓
ep073 screen raw (46:29)
Is the Atlassian.design, like the design system site, the best place for design documentation? Is this a place where designers and engineers go? Currently, yes. But in the future, this is one of the vessels of this design system information. So they've been ⁓ writing and converting all of this information for all of the ⁓ components and tokens into structured content. That is in JSON form.
Ben Callahan (46:29)
Is the Atlassian.design, like the design system site, the best place for design documentation? Is this a place where designers and engineers go currently? Yes. But in the future, this is one of the vessels of this design system information. So they've been ⁓ writing and converting all of this information for all of the ⁓ components and tokens into structured content. That is in JSON form.
ep073 speaker raw (46:29)
Is the Atlassian.design, like the design system site, the best place for design documentation? Is this a place where designers and engineers go currently? Yes. But in the future, this is one of the vessels of this design system information. So they've been ⁓ writing and converting all of this information for all of the ⁓ components and tokens into structured content. That is in JSON form.
ep073 screen raw (46:58)
typically has worked as a designer, how I've interfaced with it is I would go and call the ADS MCP and then ⁓ ask ⁓ for specific token information, for specific usage guidelines, and that's the deterministic way. The people that are working on the MCP are also ⁓ making sure that it's completely deterministic and there's no wiggle room for any sort of ambiguity. But let me try to find you a link.
Ben Callahan (46:58)
typically has worked as a designer, how I've interfaced with it ⁓ is I would go and call the ADS MCP and then ⁓ ask ⁓ for specific token information, for specific usage guidelines, and that's the deterministic way. The people that are working on the MCP are also ⁓ making sure that it's completely deterministic and there's no wiggle room for any sort of ambiguity. But let me try to find you a link.
ep073 speaker raw (46:58)
typically has worked as a designer, how I've interfaced with it ⁓ is I would go and call the ADS MCP and then ⁓ ask ⁓ for specific token information, for specific usage guidelines, and that's the deterministic way. The people that are working on the MCP are also ⁓ making sure that it's completely deterministic and there's no wiggle room for any sort of ambiguity. But let me try to find you a link.
ep073 screen raw (47:28)
That's great. you. All right, folks, we did it again. ⁓ We have a couple minutes left and somehow an hour disappears in what feels like five minutes, but cannot say how grateful I am for all of you jumping in answering the question so thoroughly. If you haven't had a chance, please go dig into that raw data. There's so many great stories in there. ⁓ Get connected with Davy and I out on LinkedIn or Blue Sky or wherever you do your
Ben Callahan (47:28)
That's great. you. All right, folks, we did it again. ⁓ We have a couple minutes left and somehow an hour disappears in what feels like five minutes, but cannot say how grateful I am for all of you jumping in answering the question so thoroughly. If you haven't had a chance, please go dig into that raw data. There's so many great stories in there. ⁓ Get connected with Davy and I out on LinkedIn or Blue Sky or wherever you do your
ep073 speaker raw (47:28)
That's great. you. All right, folks, we did it again. ⁓ We have a couple minutes left and somehow an hour disappears in what feels like five minutes, but cannot say how grateful I am for all of you jumping in answering the question so thoroughly. If you haven't had a chance, please go dig into that raw data. There's so many great stories in there. ⁓ Get connected with Davy and I out on LinkedIn or Blue Sky or wherever you do your
ep073 screen raw (47:55)
Connecting I know Davy I can probably speak for you But I love it when folks reach out and just want to talk shop always looking for that kind of conversation Redwoods is open. We've been open for a year now and I have dropped in I don't usually drop this in here But there's a code here for a free month if you're interested in joining and you you want to just test it out first You can do that here. It's so much fun A lot of folks on the call are in Redwoods we have great conversations like this every week
Ben Callahan (47:55)
Connecting I know Davy I can probably speak for you But I love it when folks reach out and just want to talk shop always looking for that kind of conversation Redwoods is open. We've been open for a year now and I have dropped in I don't usually drop this in here But there's a code here for a free month if you're interested in joining and you you want to just test it out first You can do that here. It's so much fun A lot of folks on the call are in Redwoods we have great conversations like this every week
ep073 speaker raw (47:55)
Connecting I know Davy I can probably speak for you But I love it when folks reach out and just want to talk shop always looking for that kind of conversation Redwoods is open. We've been open for a year now and I have dropped in I don't usually drop this in here But there's a code here for a free month if you're interested in joining and you you want to just test it out first You can do that here. It's so much fun A lot of folks on the call are in Redwoods we have great conversations like this every week
ep073 screen raw (48:22)
⁓ Tons of information from the folks who are over in Redwoods publishing. This is something we have a writing accountability group over in Redwoods. And so if you're somebody who wants to be writing more, wants to be sharing the things you're learning, you can join, you can be in that group and we kind of spur each other on. ⁓ Lots of fun stuff happening there. Tons of other really good articles also coming out from the community. Guy mentioned his article right here, the cost of yes, which is on the Zero Height blog. So check that out.
Ben Callahan (48:22)
⁓ Tons of information from the folks who are over in Redwoods publishing. This is something we have a writing accountability group over in Redwoods. And so if you're somebody who wants to be writing more, wants to be sharing the things you're learning, you can join, you can be in that group and we kind of spur each other on. ⁓ Lots of fun stuff happening there. Tons of other really good articles also coming out from the community. Guy mentioned his article right here, the cost of yes, which is on the Zero Height blog. So check that out.
ep073 speaker raw (48:22)
⁓ Tons of information from the folks who are over in Redwoods publishing. This is something we have a writing accountability group over in Redwoods. And so if you're somebody who wants to be writing more, wants to be sharing the things you're learning, you can join, you can be in that group and we kind of spur each other on. ⁓ Lots of fun stuff happening there. Tons of other really good articles also coming out from the community. Guy mentioned his article right here, The Cost of Yes, which is on the Zero Height blog. So check that out.
ep073 screen raw (48:51)
There's some fun events coming up. Zero Hike Converge is happening ⁓ in October in Newcastle. I'm going to be there. A bunch of folks will be. We're going to plan a little hike to go see some big trees like we do. ⁓ I'm going to be in London ⁓ for UX London at the beginning of June and would love to see you there if you're in that part of the world. So please grab a ticket. They gave me a discount code as well. Join BC. We'll get you 20 % off. So ⁓ tons of other stuff in here, folks. As always, very, very grateful for you, Davy.
Ben Callahan (48:51)
There's some fun events coming up. Zero Hike Converge is happening ⁓ in October in Newcastle. I'm going to be there. A bunch of folks will be. We're going to plan a little hike to go see some big trees like we do. ⁓ I'm going to be in London ⁓ for UX London at the beginning of June and would love to see you there if you're in that part of the world. So please grab a ticket. They gave me a discount code as well. JOIN_BC will get you 20 % off. So ⁓ tons of other stuff in here, As always, very, very grateful for you. Davy.
ep073 speaker raw (48:51)
There's some fun events coming up. Zero Hike Converge is happening ⁓ in October in Newcastle. I'm going to be there. A bunch of folks will be. We're going to plan a little hike to go see some big trees like we do. ⁓ I'm going to be in London ⁓ for UX London at the beginning of June and would love to see you there if you're in that part of the world. So please grab a ticket. They gave me a discount code as well. Join BC. We'll get you 20 % off. So ⁓ tons of other stuff in here, folks. As always, very, very grateful for you. Davy.
ep073 screen raw (49:20)
Thank you so much, man, for bringing this topic. Lots of folks, lots of interest.
Ben Callahan (49:20)
Thank you so much, man, for bringing this topic. Lots of folks, lots of interest.
ep073 speaker raw (49:20)
Thank you so much, man, for bringing this topic. Lots of folks, lots of interest.
ep073 screen raw (49:25)
Appreciate it. Thanks for chatting and I'll post some stuff, more stuff in the Fig Jam. Love the Fig Jam. Beautiful. That Fig Jam will stay open. You'll be able to read thanks everybody. Appreciate y'all being here and we'll see you in a couple of weeks for our next episode. Bye bye.
Ben Callahan (49:25)
Appreciate it. Thanks for chatting and I'll post some stuff, more stuff in the Fig Jam. Love the Fig Jam. Beautiful. That Fig Jam will stay open. You'll be able to read thanks everybody. Appreciate y'all being here and we'll see you in a couple of weeks for our next episode. Bye bye.
ep073 speaker raw (49:25)
Appreciate it. Thanks for chatting and I'll post some stuff, more stuff in the Fig Jam. Love the Fig Jam. Beautiful. That Fig Jam will stay open. You'll be able to read thanks everybody. Appreciate y'all being here and we'll see you in a couple of weeks for our next episode. Bye bye.
Ben Callahan (49:44)
Thank you so much for joining us on this episode of The Question. Remember, you can get access to the raw data, the collaborative FigJam, and all of the recordings for this episode and many more on my website, bencallahan.com If you or your team could use an outside perspective on your design system program, I'd be honored to support you in that way. There's more information about my coaching and consulting offerings over on the site.
Thanks for being here and remember, stay in learning mode.