Subscribe
Share
Share
Embed
AI First with Adam and Andy: Inspiring Business Leaders to Make AI First Moves is a dynamic podcast focused on the unprecedented potential of AI and how business leaders can harness it to transform their companies. Each episode dives into real-world examples of AI deployments, the "holy shit" moments where AI changes everything, and the steps leaders need to take to stay ahead. It’s bold, actionable, and emphasizes the exponential acceleration of AI, inspiring CEOs to make AI-first moves before they fall behind.
Niel Robertson (00:00)
So I was like, okay, I'm never using software again. So I realized that like I
want to provide this ability to abstract all of the complexity of UI because the UI is in this we can go on forever about this. in the software development world with AI, the limiting factor now is designing UI for humans. That's the thing that takes the longest amount of time to do now. I wanna get away from
Adam (00:18)
Sorry, yeah, what what sorry, can you say
it again? What takes the longest amount of time?
Andy (00:21)
Designing
for humans.
Niel Robertson (00:23)
Yeah, designing for humans. That's the hardest
part of the software engineering pipeline now. It takes the most amount of our time. Everything else, we've got massive amounts of AI tooling, but we sit there and we tinker with the UI for humans because you're dealing with multiple things like that.
Adam (00:33)
Right. That's so true.
Niel Robertson (00:35)
And I'm like, I don't anymore. I just use that literally, I don't touch software anymore. I go get MCPs and I use the MCP because I can have a simple interface which is conversational.
Andy (00:44)
This is AI First with Adam and Andy, the show for leaders turning AI from a tool they're testing into a workforce they run. I'm Andy Sack, alongside my co-host Adam Brotman. Each episode we bring you candid conversation with business leaders deploying real AI agents. What's actually working, what's not, and how they're rebuilding their companies around it. No fluff, just real talk, actionable use cases, and insights you can use.
Today's episode, we have a guest, an old friend of mine, Neil Robertson, from Winslow. And we're gonna get to him in just a moment. but what you'll be getting today is really an upfront conversation on
HR and how the organization, is changing, how the tools for HR is changing, and what AI and agents mean for HR and the org. With that, Neil, welcome to AI First with Adam and Andy.
Niel Robertson (01:40)
Thanks guys. I'm really excited to be here. So we just just launched our product a couple weeks ago. So I'm very excited to talk about what we're doing.
Andy (01:47)
Neil, give a brief background on you, but then talk about Winslow and what you launched.
Niel Robertson (01:52)
Yeah, I mean, so Andy, you and I know each other because I sort of showed up in the early nineties in the Boulder Colorado startup ecosystem, started hanging out with everybody, built a company there that ended up being kind of a right time, right place, first wave internet company. and continued on as a sort of a venture-backed entrepreneur.
Over a number different companies, worked at three different venture firms along the way as kind of a venture partner, s have sat on sort of both sides of the table, done a lot of work with startup communities, startup programs like Techstars. I ended up helping bring Techstars to Australia here in the early days. and have had just a passion my whole life for
being a product builder and also being a company builder. my personal background, I went to MIT, I have a C S degree from MIT. At various points in my life, I've either run companies or written software and now I do both again, which is really fun. Although nobody writes software anymore. So I tell agents to write software for me. and what's really fun about this moment in time is both of my itches get scratched. I get to build new and interesting products, but I also get to build companies in a totally different way. and it's a fun kind of like duality every day. So
That's my background.
Andy (02:53)
All right, so
tell our audience about your recent venture and what the tool does.
Niel Robertson (02:58)
Yeah, so Winslow is a complete rethink of the traditional payroll benefits HR and operations software stack. People have been using the same software that was built between 10 and 20 years ago, kind of like the Gusto's Rippling's workdays of the year to do these sort of like tactical operations. And you get very quickly dragged into kind of a old school hyperclerical.
Non-scalable ways of doing things. AI completely revolutionizes the ability for people to use and run these pieces of software in a much different way. So Winso is a completely AI native version. I mean it's 100% conversational, it's a hundred percent agentic, it's a hundred percent extensible. and also it sort of pairs with kind of a next generation piece of functionality, which is kind of an AI coworker inside of Slack. Think of it like your chief of staff.
And so not only are we kind of reinventing the core ways that traditional sort of payroll benefits software works, but also rethinking how that software can be meaningful to every employee in your business. One of the things I think is sort of crazy about the industry is HR software is one of the few pieces of software that every single employee has a site license for or has a license for, and almost none of them use during the year. And so you pay this exorbitant amount of money for software that everybody hates but doesn't use.
We think it should be completely the opposite, where people are using their HR software on a daily basis to become more productive. And like we've introduced a product.
Andy (04:17)
And who's the customer of the product?
Niel Robertson (04:19)
So we serve two primary customers. in general, we serve kind of like one to five hundred employees-sized companies. That's either first payroll, so that would be usually a startup CEO. He or she just got a little pre-seed money and wants to like do things in an official way, they need a payroll company, they would choose us over like a classic legacy vendor like Augusto or Rippling.
We also work with companies that are starting to scale and they might have used a company like a PEO, like a Just Works or a TriNet, and it's getting way too expensive to have that kind of service provider anymore. They want their own software and they're coming to us at 50 people or so because they know they're gonna scale to a couple of hundred. these days it's a fractional HR person. Almost no one hires an HR in-house person anymore until you're much bigger. So it's a fractional HR person. Like we could talk about like the the growth of fractional is like
of the ecosystem, it's like really interesting. so it's either like a CEO or usually a fractional HR person that's implementing.
Andy (05:13)
And and generally
smaller companies, not not big enterprise.
Niel Robertson (05:17)
Yeah, zero to zero to five hundred, probably more densely in the zero to two hundred, sorry, in the one to two hundred and fifty employee range. companies complexity sort of like grows in step functions. And once you go kind of above 500 employees, a whole bunch of weird stuff starts to creep in. And it just makes the necessity of the platform a lot more complex.
Andy (05:33)
So I wanted to provide our audience a little bit of context. I mean, Neil is an old friend and a great entrepreneur, builder. the reason that this episode came to be is we just launched Agents Inc. As part of Agents Inc, there was a lot in there about the HR function, the changing nature of
Companies
and organization, as well as predictions about what's going to happen to labor. And as Adam and I spoke about it.
We thought that would be really helpful to have someone who was sort of an expert in HR and I thought of Neil. so with that context, I want to invite Adam into the conversation just with whatever question he has for Neil in the domain. Like, Neil's a builder, he's building for HR professionals. You heard his customer base, like tee us off on the more relevant conversation for our audience as it relates to HR or design, labor, wherever you want.
Adam (06:24)
Okay, yeah, Neil, super interesting. I want to double click, so to speak, on your comment about conversational versus the way this software has been built for the last 10 or 20 years. and my question for you would be: first of all, just to kind of set the stage so I understand, but I'm wondering if our audience has the same question, which is like, okay,
Today, if I'm running a 50-person company and I don't really have an head of HR, or I have one person that's head of HR and they're doing everything, they're using a gusto or something.
Niel Robertson (06:53)
Mm-hmm.
Adam (06:53)
do you come in and basically say, like, yeah, instead of like deterministic software that captures information, puts it in a database somewhere, there's kind of like
point click file menu kind of interface for everybody is is
Niel Robertson (07:05)
Mm-hmm.
Adam (07:06)
yours more we're gonna do the same thing that a lot of these other systems are doing but instead of point click it's just through chat and then they sometimes get like HTML emails they sometimes get stuff that happens through the chat but it just kind of explain like from what to what a little bit more in a little bit more detail.
Niel Robertson (07:22)
Yeah, let me give you a number of examples. So a way to think about our product is we've designed the product both for a sort of a die-in-the-wool HR person that might want to go in and use a user interface they're familiar with. So they might want to go do a payroll review and a pretty classic payroll review mechanism. But then also you've got this 100% conversational interface, which uses our underlying MCP tooling. And so
For those that don't know, you know, MCP tooling is the sort of way that agents talk to software now and figure out what they can accomplish. we have the most robust MCP tooling of any company in the industry, like from kind of like workday to modern companies. And let me give you an example of like why that becomes powerful. Like, let's say that you are actually a three-person company, you're the CEO, you got to do payroll. There's still like just because like the software changes doesn't mean you don't need to understand all the regulation and complexity of it.
And so say for example that you are hiring somebody new and in their offer letter you decided to agree to give them a hundred and fifty dollar gym membership. Or in California, you know, you need to pay by law sometimes $50 of a internet and phone supplement to certain types of people.
How do you get that into the payroll system actually in the right way? There's actually a big difference between pre-tax and post-tax deductions inside of payroll. And you have to kind of know categorically which thing you're dealing with falls into each category. An HR person who's been doing this for like 10 years would just know that because they've done it a hundred times. A first time CEO probably doesn't know or doesn't care. You go into Claude, which has all of the intelligence of the universe, and say, Here's the offer letter, and I need to put them into Winslow.
And it goes and it negotiates with our MCP tools, but then also goes and looks at the intelligence out on the internet to understand, you you added the gym membership. That is a that is a pre-tax deduction. I've got to put it in as an adjustment, not a reimbursement. And it negotiates with our system to get all that stuff into the database correctly. So it's not simply about the user interface and reducing all the complexity of trying to go find the right places. It's actually about
The combination of user interface with domain knowledge, that's very hard for someone that's not an expert to sort of navigate a traditional HR system and get things correct, because you can make these mistakes and they end up getting fined and penalized for them. That's kind of like one way to think about it from a simplistic perspective. A, you could work in Claude or you could work in Slack and get that done, but you could also hand a lot of the domain knowledge to Claude to figure out how to negotiate with our MCP to get correct. Does that make sense?
Adam (09:45)
Yeah.
You know what it makes me think of as you're describing it, like one of the interesting, like almost like unlocks of using AI for what you're describing is that use your example. I'm a CEO with no domain experience or knowledge by HR. I don't have a separate HR person. there's both
The domain knowledge of like a functional leader of HR, and then there's like the software to manage stuff. because of the way you're describing this, I'm thinking in my head, tell me if I've got it right, you're sort of combining those things. You're like, well, why can't the CEO or the HR leader, instead of having a staff, they can basically go to this AI-powered HR software, and it's like a combination of doing what the deterministic
SaaS software systems would do as well as what maybe somebody on their staff would do. And it's like a little bit of both at the same time. Is that right?
Niel Robertson (10:35)
Yeah. But I've had a couple like significant epiphanies in the last six months that are for my view of the world. If you'll indulge me to give them to you. So one of them was I was talking to someone and they needed my 409A like fair market value. And I was like, Okay, it's in Carta somewhere. I don't want to go digging around in Carta. I've done it before, I gotta find it. So I went into Claude and I said does Carta have an MCP? And it's like, Well, of course it does.
So I just added the MCP and I said, go get my last FMV out of Carta. I never went into Carta. I never navigated their UI. Whether the UI was the winning design awards or the worst UI in the world, it doesn't matter to me anymore. I was able to extract the thing I wanted accurately. And I don't care that it's running asynchronously in the background and took two or three minutes to do it. That's fine. So I was like, okay, I'm never using software again. So I realized that like I
want to provide this ability to abstract all of the complexity of UI because the UI is in this we can go on forever about this. in the software development world with AI, the limiting factor now is designing UI for humans. That's the thing that takes the longest amount of time to do now. I wanna get away from
Adam (11:36)
Sorry, yeah, what what sorry, can you say
it again? What takes the longest amount of time?
Andy (11:39)
Designing
for humans.
Niel Robertson (11:41)
Yeah, designing for humans. That's the hardest
part of the software engineering pipeline now. It takes the most amount of our time. Everything else, we've got massive amounts of AI tooling, but we sit there and we tinker with the UI for humans because you're dealing with multiple things like that.
Adam (11:51)
Right. That's so true.
Niel Robertson (11:53)
And I'm like, I don't anymore. I just use that literally, I don't touch software anymore. I go get MCPs and I use the MCP because I can have a simple interface which is conversational. That was like one big epiphany I had. The second big epiphany I had was
About this understanding of the movement from kind of like first generation conversational functionality in SaaS software to second generation. I was using Linear to do something to put a ticket in. And I opened up their chat just to kind of play with it. And I said, Linear, can you put in this feature request that I have? And do me a favor, go look at what is publicly available about Gusto and Rippling and do a feature analysis and put that into the ticket.
And linear went, what's A gusto? And I was like, okay, that's interesting. that's V1 of chatbots, where they don't understand the world. V2 of chatbots, which is how our product works, has all of the knowledge of Claude available to it to do anything. So if you say, Hey Winslow, go do something, and I need you to understand deeply about like cobra regulation inside of California, that is part of your product.
Like you have basically taken what I call the infinity machine and you've brought it
Adam (12:58)
Right.
Niel Robertson (12:58)
into a product. And like there's a that's a big difference between architectures of AI native software that think about it's our software, our system of record, plus everything that Claude knows, and being able to bring that to bear for you because of how we've architected our system versus older generation chat products like HubSpot's Breeze or Linear or whatever, who are kind of like locked into this world of like they only understand their own product.
And once you like start to realize next generation, like I can go into cursor, for example, if I'm coding something and I can say, hey, I want to build this feature, go take a look at four payroll systems out there and make sure that I'm adding the right functionality. And it knows how to do that because it's using the infinity machine of Claude or Banana or whatever. Those two things form my view of the world. And in the HR world, it's very important because HR is a domain of subtlety. So for example, like
I was just going through a termination process and an employee in California or a Colorado company and they'd only been with the company for a little bit. So you have to abide by like Colorado like mini Cobra laws in terms of they have Cobra available. There's like all this massive amount of subtlety that goes into just like hitting the termination button. That can now be part of my product because Claude knows how to interpret all of that, talk to my MCP and configure everything correctly for me. So there's no amount of UI that I could build in the traditional sense that would have gotten that amount of
Adam (14:13)
Right.
Niel Robertson (14:13)
functionality
into a startup CEO's hands. And now I can, which completely changes the way that I design software, knowing that like the infinity machine of knowledge is sitting there and I can connect it to my product, if that makes sense. It's a very, it's a completely not only game changing
Andy (14:28)
Yeah.
Niel Robertson (14:28)
way to think about how software works.
Andy (14:30)
While you're talking about your product, let me just name it.
a lot of our users will not know necessarily what linear is, MCP. right, like your Neil as a guest is a MIT educated CS engineer with founder, et cetera. He's like and he built products. So forgive us for the technical talk.
Niel Robertson (14:49)
Yeah, sorry. Yeah.
Andy (14:49)
That said,
you said something at the beginning, Neil, which was about HR software, I want to continue to take advantage of your expertise, which is HR software is this unique thing, which is people pay oodles of money for and a site license for every employee, and people log in once a year and that's it. How does Winslow?
change that and you had stated at the beginning which was good marketing banter maybe it's marketing banter maybe it's real how do you actually make it real in your product
Niel Robertson (15:16)
Yeah. So two things. One is we threw away the traditional per employee pricing model for HR tools and we have a usage based model. And so that's a big step in the industry. So
Andy (15:26)
Yeah, no, that's big.
Niel Robertson (15:26)
yeah, most of the industry, the work days of the world are propped up because they charge, you know, seven to twenty-five dollars depending on the number of modules you have per employee per month, regardless of people logging into their software or not, right? Just a huge amount of that industry is just like wasted on software.
That's being spent and no one's using it. So we have a complete usage-based model based on how much you exercise our AI technology. That's a tactical solution to the problem, but here's the real solution to the problem is it turns out that many things that people would do to make their lives more productive as employees involve employee data. So for example, let's say that I am the CEO of a company and I really want to welcome every new employee to the company because we're growing very quickly. And so I want to send them
a Slack message on the second day that they're employed, say, hey, welcome. I'm so glad you're here. Here's my Calendly link. Why don't you grab some time with me, 30 minutes on Zoom? I'd really love to just have a chat with you, get to know you. And then if they like forget to do that, you want to like send them a reminder in a few days. And then if they don't get around to doing it, you want to ping their manager and say, hey, can you make sure this person does it? You can go into our Slack bot and literally say that.
When an employee starts on their second day, send them a DM, send them a calendar. If they don't respond in three days, send them a reminder. If they don't respond in seven days, send a manager a reminder. You hit go, and we scaffold up everything agentically in the back end to do that for you. So if you kind of know behind the scenes what's happening there, you're listening to events from the system, you're managing scheduling, you've got skills files that are getting built. We all do that completely automatically in the back end, which is what I called our chief of staff. but that
piece of productivity that I want, it needs to know about managers, it needs to know about start dates. And so if you dig through all of these things that people really want to do with, for example, data, if you're in the sales team, you want to say, hey, I want to see how long it takes for my sales team to become productive. So go and
Take a report of all of my closes and also get a report from Winslow of like start dates, put them together and just build me a graph of like how long it takes for people to start closing. That's employee data from our database and sales data from HubSpot or ADIO or whatever you have. There's so many things that make your job more productive that require a little bit of knowledge that comes from the HR system that you don't even touch now because you can't get access to it.
So when you give people these tools, it unlocks this massive amount of creativity, which creates productivity in the system. And all of a sudden, the HR product is something you use every day, not something you use one, two, three times a year. That's the unlock that I realized. And as we started to do it for our own company, you're like, this is amazing. The other thing that we did in our business is HR systems are traditionally read-only. You cannot build a product that writes into them.
Our product is writable. So you can go and create like let's say that you build your own performance management tool in lovable. You're like you just want to go vibe code something very specific to your company. You can go build a performance management tool, and then with our product, you can actually push the results of the performance management of an employee onto their employee record. It's stored in Winslow on their record, and now that becomes queryable.
So CEO could say, I have a meeting with Andy in a couple of days, pull up his last performance review, and it will just go get that out of the employee record. No, I've got to go into Lattice or a third-party system and pull that down and merge it together with the record. You want your employee data to be like a CRM system. It's sucking in all of the digital exhaust for everything to do with that employee and the work they're doing. So you can do analysis and you can build tools on top of it.
It's a whole mentality of how to think about the power of an AI native architecture inside of an organization and the unlocks that you get to make people more productive.
Andy (18:48)
Neil, given you're building
tools for HR professionals, predominantly of smaller companies, can you talk about the
Niel Robertson (18:54)
Thank you.
Andy (18:57)
changes and stresses on HR professionals in the last 12 to 24 months and things that have surprised you just about the job and the market and how it's changed?
Niel Robertson (19:07)
Yeah. so I think AI has been a challenging technology for HR professionals for a few reasons. One is it's generally not a technical organization. and this has actually been one of the things that's really unfortunate about HR for a long time because it's not a technical organization. They've kind of had to go hat in hand to IT all the time and say
I really need an integration done. I really want this thing built. You know, they're kind of very deferential to what IT lets them do.
And they're kind of in many cases not as empowered as they could be. The thing that's amazing about modern tools, AI native tools, as I mentioned, were 100% extensible, they can build anything they want now. They don't have to go ask IT for anything anymore. So there's this moment where all of a sudden the lack of technical depth is not actually that big of a barrier anymore.
And the platforms allow them to actually start building what they want. So our positioning in the market is build HR the way you've always wanted. It's an empowerment approach to HR.
Andy (19:59)
So now let me interrupt,
let me interrupt you for a moment. I'm trying to get you off of Winslow and to comment more
Niel Robertson (20:04)
Sure.
Andy (20:05)
on the market and the people. I'm just aware of who our audience is. I'll put this last plug-in, which is if you're thinking about using needing or wanting to empower your HR professionals, go to usewinslow.com.
Check
it out. Now I want to talk about three things in the last 15 minutes. One is HR professionals and how their job has changed as a result of agents and AI. Because like the workforce is fucking changing. Excuse my French. Let's talk there. The
Niel Robertson (20:31)
Mm-hmm.
Andy (20:32)
second thing I want to talk about is org design, and the third thing I want to talk about is
the fear of employees because of AI and agents, which changes the job of the HR professional materially, that fear. So pick any of those
Niel Robertson (20:44)
Yeah. Okay.
Andy (20:46)
and and opine on that and not Winslow the product.
Niel Robertson (20:49)
Yeah, okay. So the things are all commingled, right? So let's take organizational design. I think in general, what we're seeing emerge is what I call the triangle design, right? Which is historically you've had the head of department and underneath them you've had the way that stuff gets done, which is people.
Right. And then over here you've had some kind of operations role, right? So let's take HR, for example, you've had the HR CHRO over here, benefit specialists, payroll specialists, HR generists, HRBPs, like tons and tons of people that kind of get tactical things done. And over here you've had like sort of peep-lops working around for second. This is being gobbled software. This is all AI. So slowly but surely, all of the how things get done is software, intelligence, agents, etc.
What it's doing is it's causing the people ops role to become very, very, very important. So you see the people ops role moving much into the foreground where it used to be in the background. And that is because it's a people ops are traditionally a combination of technical.
They understand business, they understand people, they understand workflow. That's like the premium role. You're seeing this in all organizations now. We think the same thing in sales, right? You've had sales, you've had salespeople. This is being gobbled like AI, BDRs and SDRs, but now the sales ops role, it's very, very important. You gotta glue it all together, right? Same thing in marketing tech, same thing in fintech, same thing in DevOps. It's the exact same triangle that's happening. So HR teams are sort of like trying to figure out how to like reorganize around these sort of triangle structure.
Where the ops role becomes much, much more important. The reason I bring up the technical is because HR traditionally has been an organization that hasn't been able to move that independently. They've really required other parts of the organization to do things for them. That's not true anymore with AI. They have the ability to build things themselves. That actually empowers them in a totally new way to go and construct.
what they want. They've also been pretty limited because these big monolithic platforms that have existed for a long time have kind of forced them to do things the way that the monolithic platforms wanted them to. And so they've been kind of constrained by their IT department, they've been constrained by these platforms. And they haven't just been able to build what the organization needs because they don't have the technical skills. Not true anymore with Vive Coding Tools. Independent of our platform, you don't, you know, you can still go do this independently. This is a really big sea change for HR Teams where all of a sudden the things you couldn't do and the things you
To beg for before you can now do like it's amazing. They woke up and had a whole bunch of toys under the Christmas tree. It's like great, right? So they're trying to learn those things and adjust those things and figure out how to do change management on those things. You know, one of the downsides, for example, of vibe coding tools is people get excited, they start building a bunch of stuff, and then A, they have to maintain it. You know, obviously there's a lot of storyline about that, but also like all of the data that those things generate are sitting in a bunch of different places now, and they don't have it centralized anywhere.
So they're starting to grapple with things like centralization of data, cleanliness of data. There's all these kind of very like practical, tactical things that happen once you start to empower an organization to do things in a consistent way with AI, because AI needs a source. And so they're kind of getting sucked into a lot of like cleanup work that they have kind of avoided for a long time.
It's like, you know, if you build a sales organization, at some point in time you go, Salesforce is mess. We need to go clean up all this data because we've got like all sorts of different stuff lying around and we can't run reports really well, and it's all coming out kind of wonky. It's the same thing in HR. You give them a bunch of AI tools, they get super excited, they go build a bunch of stuff, and then you realize, okay, our house is kind of not in great order. We have to stop and kind of go spend time and energy on that. That's what we're seeing a lot of people grapple with. And then HR teams are really struggling with.
The flow of very sensitive information through AI systems. So for example, I could go into any AI tool and any product, whether it's ChatGPT or Glean or Linear or whatever, and I could type in: I just got diagnosed with stage five cancer. I live at this address. I need to find an oncologist near me. I've pumped huge amounts of PII into a chatbot.
And how does that information flow through the supply chain? Where does it get stored? So, rightly so, HR people's hair is on fire a little bit about data protection, compliance, supply chain stuff. And so
The thing that I find, which I think is challenging, is I think sometimes they let those issues get a little bit in their way of making forward progress. and they kind of need to find a balance between being like hyper conservative about all the risks there and also taking advantage of the opportunities to kind of operate in a different way. So I mean, is that kind of like thing that you're thinking about? I mean, that's what
Andy (25:04)
Yeah. Yeah, yeah. Adam, you have anything you wanna
chime in or ask here?
Adam (25:10)
Yeah, I mean it's interesting. I'm still on Neil's epiphanies, by the way, because but I'm
Andy (25:14)
Yeah, yeah.
Adam (25:15)
gonna apply that at a broader level. like what I think is so interesting about what Neil is saying, and I think it's ironic slash interesting that he's talking about it from an HR perspective. but you said it, Neil, it could apply to a lot of functions, which is that like
Andy (25:27)
Adam, just
to get which piece of his epiphanies are you referring to?
Adam (25:30)
Well the two
Epiphanies, as I understood them, maybe I'm sort of blending them, but
in the triangle organization as you kind of put it, but in general, you you used to have like data that was housed in systems of record software, that somebody's job was to understand how to access that system of record software as part of their overall job. They reported it up to a broader organization that at times, either within that function or in general, would need that person to go access that system that housed that data.
And what Neil's epiphany was is like there's actually two layers there that are not necessary in the age of generative AI theoretically, which is that you don't need a person to then go interact with a deterministic software interface that's housing data. in some senses, you can go straight to an intelligent agent yourself in a conversational manner and
skip two of those steps. I don't need the person and I don't need the system of record. I need the data. Okay. I need the data and I need it in a read-write. I need it secure. I need it organized in with a semantic layer, but I don't need the person or the software layer that I used to need in a sense. Because I can just go if you think about all that is, is that there's software that's allowing the data to be accessed, invisible, and queryable.
By a person that knows how to use that software. But an agent or an intelligence system, it just needs a connector into either that system directly, which eliminates the need for the person, perhaps on some level, but it doesn't
Niel Robertson (26:56)
Okay.
Adam (26:58)
need the interface either, like Neil said. The other epiphany like so much of our stuff is being designed for a human. It's like, no, just give my intelligent agent access to either the system of record or frankly,
decompose
Niel Robertson (27:09)
Yeah.
Adam (27:09)
a layer below that to the actual system of records all that system of records is doing is like housing, organizing, and giving access to that data. So, and the intelligence
layer itself not only can do all that, but it's not just an order taker, like, I used to just go tell a person to go get me this thing. It's actually gonna be better at helping you get to why you needed that data in the first place anyways. It's like, what am I trying to accomplish with that data? I'm
Really trying to figure out this insight over here. I'm trying to do whatever. And that's so I just was like struck by Neil's point. He's talking about Winslow in some ways in the epiphany, but I actually think it's goes right to the heart of what we've been saying, Andy, and what Neil is saying too, which is like, the pathways that we've all built in terms of human organization and technology organization are gonna be completely rewritten and have to be unlearned.
Niel Robertson (27:54)
Well, like I give an example, like we use HubSpot and like all kind of systems like that end up with a complicated reporting product, like a dashboard product. It's the nature of every single SAS product ever. And I beat my head against the wall every time I need to go and try to get data out of HubSpot because their reporting interface is very complex, sophisticated, use whatever word you want. Not anymore. I attach their MCP and I just say what I want and it gets it. And so I literally was doing what you were saying before, Adam. I was like, I was working with
a fractional marketing person who knew Hubspot really well, I would tell her what I wanted. She would go into HubSpot, she would create the report, stick it on a dashboard for me. And I was like, this seems insane. And so now I'm like going to go into Claude and say I want this piece of data and you can schedule it and send it to me every morning at three AM into my Slack channel. I have so much of that stuff in Claude.
Adam (28:38)
Yeah, Neil, just to build on that for a second, which is what's so interesting to me about you're saying is A, if you're right that the thing you wanted was exactly the right thing to go gather,
Niel Robertson (28:49)
Mm-hmm.
Adam (28:50)
great. we've just talked about how why do I need to go ask this person who knows how to use HubSpot to then go get this thing out of HubSpot? I can just go to my agent and ask my agent with the connector to just go do that for me and format it in way I want, right? But
Niel Robertson (29:02)
Yeah.
Adam (29:03)
There's another benefit of using an agent, which is like, are you sure that that's even what you needed? I mean, probably is, but maybe
Niel Robertson (29:08)
There you go. Yeah.
Adam (29:09)
not. if you start with being an output manager of like, what am I trying to accomplish? Why did I even want that data in the first place? And you go to the agent and say, I'm trying to accomplish this.
Niel Robertson (29:18)
Mm-hmm.
Adam (29:18)
my normal thing would be to go ask somebody to go get this at a HubSpot. you get the benefit of an agent that says, well, actually, if you're trying to accomplish that, there might be a better way to go get that insight.
And I'm gonna go get that for you. Or you might be right and I'm just gonna skip the step. Either way, you're like faster and better off using an agent in this. And it's crazy that we're talking about this in a world where none of this was possible a year ago.
Niel Robertson (29:40)
I have a friend that works at NVIDIA, a very smart guy. And he and I call this planning mode. and planning mode is like when you're working with someone or talking to someone and they're having you qualify a lot of the questions about the thing you're asking to do. And I think like this is maybe a way to articulate the second epiphany, which is I
Might say, hey, I want to onboard this new employee, make sure there's all this information in the offer letter. but it's actually not the HR system that's saying, Did you know that a gym membership is classified this way? Did you realize that the sales commission should be set up this way? make sure that you get this information to the bank account because of this time frame process that has to happen for pulling on this payroll. It's actually Claude that's doing
And so like it feels like our
Adam (30:23)
First night.
Niel Robertson (30:24)
product or an HR product or whatever is very, very, very smart, but it's actually Claude that's very, very, very smart. But our product has enabled us to work in tandem with it. And like that's the unlock where you're like, any
single piece of software
Andy (30:35)
So
Niel Robertson (30:37)
you use has the infinity of the knowledge of the universe at its disposal if you've built an agentic version of it where it can kind of go back and forth like that.
So like our product doesn't need to like have a wizard in it that says, you know, do they have a, you know, a severance payment? Do they have a this? Okay, it's severance in California, it needs to structured this way. Claude is just sort of like interrupting you, trying to get something done and being like, by the way, there's all this stuff you need to know. Can you tell me a bunch of your choices? And then it's negotiating with the MCP of the underlying SaaS product to figure out how to get that into the system of record. And like that's
Adam (31:09)
Andy's dying because Andy's
Andy (31:09)
Mm. Yeah. I want I want
I
Adam (31:11)
dying
because he he didn't want this conversation to go this way at all. Instead, you and I are just nerding out on like the possibilities of this system. He wants to talk about like how many people are gonna
Andy (31:16)
Yeah, yeah. That that's okay. That's okay. Let me break
Adam (31:20)
lose their job and what the unemployment rate's gonna be next year.
Andy (31:22)
And let
me let me go let me
Niel Robertson (31:23)
Okay.
Andy (31:24)
go back because we're running low on time. I'm interested in this topic too. It's just I think our audience might not be. the Yeah. what I
Adam (31:30)
He thinks that, but he usually they are, so it's okay, Neil. But go ahead, Eddie.
Andy (31:35)
w what I wanna pick up on is what you were saying before, so I'll bridge the gap, which is you had said, you know.
Employee's sensitive information is now being sucked into your product and going through an MCP server. And that changes the HR professional, the human professional's job. And there's some employee concern of that. I want to sort of go back to the level of fear around.
that we're seeing in enterprises and of employees about being laid off because of AI, which totally changed like HR professionals now need to deal with that all the time when they're told to interact with new software and
At the same time, I might be more comfortable talking to AI Claude about what's going on in the workplace than I am the HR professional. like it's
Niel Robertson (32:26)
Yeah.
Andy (32:27)
relatively unbiased, and yeah, my PII data might be being leaked out, but there's a lot of questionable.
Perhaps amoral, unethical things that I don't know about or know how to deal about as an employee that I'm having to make choices and I might feel more confident talking to Claude than I am through Winslow, than I am the HR professional. are you seeing any of that? Like
Niel Robertson (32:51)
Yeah, all of the data and the surveys out there actually point to exactly what you said, which is many times humans will bring up tricky workplace issues with AI because it's dispassionate, has no dog in the fight. I mean, maybe deploy
Andy (33:02)
It's safer.
Niel Robertson (33:03)
Yeah, say for maybe an employee heard that like, you know, their friend went and talked to HR and didn't have a great experience. They don't want to have that, you know, there's a lot of that stuff that goes on. Even HR professionals will tell you that this is what they see as well. but that doesn't obfuscate the need for HR professionals. I think the problem is that the HR profession has had like the most important parts of it buried in clerical work.
So you have these people that join the industry because they they want to help humans, they want to help organizations, they want to help, you know, set strategic direction. And then their inbox of just like, hey, do we have president's day off? Hey, can you help me figure out what my deductible is? Is so deep that they never ever get to the core nuggety, noogety golden stuff. And so the
Positive promise of AI is like have AI do all the clerical stuff. And all of a sudden you now get to work on the stuff that you actually really care about. And that I think unlocks a massive amount of opportunity and potential for HR to be a driver and a contributor to organizations in a way that it hasn't been before because it's been so very busy just like dealing with tactical BS. I will also say that I think personal opinion, we are in a moment in time where.
it feels like there's a contracting force and an expanding force. And the two things haven't rationalized yet. The contracting force is AI can do a lot of these tactical clerical jobs. I mean, what I mentioned in the like how stuff gets done category, you know, a payroll specialist, you know, benefit specialist, a lot that stuff can be done in the end by AI and will be done in the end by AI.
There's a contracting force, clearly. This is true in every single department. But there's an expanding force, which is, as you guys know, the surface area that you can attack as a company now because of AI is increasing really, really heavily. And as that starts to happen, the opportunity to plug yourself into an organization and be valuable in there grows with it.
And like those two things have not rationalized yet. So you have a contracting force that people perceive and get scared about, but they haven't seen the benefit of the expanding expansion force yet. You see this in small companies that are doing like a hundred million of revenue with 50 employees. Like you see that happening, but it hasn't really manifested itself in larger companies yet. But I think it will. And once it does, people will feel more confident that it's not.
about mass layoffs and the apocalypse here. It's about renavigating in the organization to figure out where your skills can impact that expanded surface area. So I'm
Andy (35:19)
Yeah, that's all said.
Niel Robertson (35:19)
like less worried about, you know, the economic drama in the next five years. I just think there's a little bit of a tale of two cities going on right now. And like we'll figure out that it's actually the same thing and like where the connection points are. That's what I think is going to happen. Like
Andy (35:36)
Great.
That's great. With
Niel Robertson (35:37)
Yeah.
Andy (35:38)
that, I'll turn it to Adam. Adam, anything you wanna highlight for our audience as we close out?
Adam (35:45)
No, I mean Neil, this is great. we haven't really talked a lot about, you know, systems of record and how they can be sort of rethought through in an agentic way. And I thought that was a cool way to introduce it to our audience.
Niel Robertson (35:57)
Yeah.
Andy (35:57)
Yeah,
I mean it was not the conversation that I anticipated having. I do think that
The epiphanies that you shared and the evolution, really the the disregard for user interface to software products and being able to throw that away in the trash and like everybody that's listening to this podcast can relate to particularly on government websites. Where the hell is that button? You know, and like and just how hard it is to build products that feel intuitive. I mean it
was Google's invention. Here's the box and here's the button, right? Like we're stripping everything else away. The fact that that is all gone and we're now interfacing in a conversational matter sometimes often in the new case via voice, we're in a whole new world. And I think you gave frontline views of those epiphanies to our audience and thanks for joining. And I think they'll take from it what they will.
with that
I wanna say thank you to Neil for joining. he dialed in at 7 a.m. from Sydney, Australia, to join this podcast. So thank you, Neil. And thank you to our audience for listening to AI First with Adam and Andy. For more info on how to become AI First, head to Forum3.com. You find case studies, research briefings, executive summaries, and our new book, Agents Inc. How to Rebuild Your Company in the AI Age.
And there you can meet Vera, our AI agentic co-author, who's always up for a conversation. If you want to connect with other leaders doing this work, join the AI First community, our hub for turning AI hype into action. Onward.