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.
Adam Brotman (00:00)
I also want to call out the confusion that I think a lot of us have. I mean, us being those of us that think about and study AI and those of us that don't. Like, I agree with Aaron just said, I want to kind of point out why I agree with it. What Aaron is basically saying is that
The concept of what is agentic or what is an agent is a spectrum. But it definitely, almost by definition, has to have some kind of tool use. Even if it's like, how do I put this? Even if it's just going and doing some Aaron in the tool that you gave an example of, like, go do some research for me, the tool that it's using is internet search.
Right. So the fact that it has its own ability to do internet search and not just answer a question from its training data. Right. Or from or an attachment I gave it. So it can go outside of its current context window. Right. And go pull something into its context window and do something for me. In the one case, it's going into search and then compiling that, thinking it through and giving me an answer about a travel trip. And the other example you gave, which I was a good one, is I'm going to give it access to my inbox.
sent and inbox both, and I'm going to ask it to do something for me I'm giving it now in both cases, I gave it a task, not just a question, and it's completing that task for me with some tool use. I do feel that that is a form of agentic behavior on the part of the AI system. I went from question and answer to task, but as the task becomes
more complex with more tools and longer running and starts to involve things like memory, proactivity, scheduling. Now you're getting into like, it starts to feel like a co-worker.
Forum3 (01:40)
This is AI First with Adam and Andy, the show that takes you straight to the front lines of AI innovation and business. I'm Andy Sack and alongside my cohost, Adam Brotman. Each episode, we bring you candid conversations with business leaders, transforming their businesses with AI. No fluff, just real talk, actionable use cases and insights for you.
Super excited about today's episode. We're continuing down the path of our exploration of agents in enterprise with a specific focus on agentic marketing today. And our guest happens to be a domain expert, ⁓ also a friend of both Adam and I. Allow me to introduce Aaron Bird. Aaron, welcome.
Aaron (02:32)
Thanks, Andy. Yeah,
excited to be here.
Forum3 (02:34)
Aaron, do want to just for the audience, just I think introduce yourself first, but then also your company and what your company does. And then we'll get into the conversation on agentic marketing.
Aaron (02:45)
Yeah, so I'm Aaron Bird. I'm the co-founder and CEO at Inflection, inflection.io. We're AI native marketing automation. We focus on B2B, so our customers are all B2B. And I've been building in the MarTech space for maybe like 15 years.
Before this, I ran product at Marketo, a ⁓ legacy B2B marketing automation provider that Adobe now owns. Before that, I was the co-founder and CEO of a company called Visible. We pioneered B2B marketing attribution, so marketing measurement for B2B companies and tying marketing activity to revenue. We sold the company to Marketo.
And that's how I transitioned to run product there. then long ago, I was at Microsoft in the ads platform there. I've been in like generally in Martech for a while and honestly super excited about the phase we're in. I've got enough gray hair to remember the dot com boom. I was kind of graduated from college back then.
This feels just like the same thing, like the transformation that happened with the internet. I was a very young person at the time and kind of watched all that happen and this feels identical. It's just moving so fast and it's just so fun to be building right now and I'm excited to talk about that.
Forum3 (04:18)
Awesome. Adam, do you have a question you want to tee us off with with Aaron?
Adam Brotman (04:23)
Yeah, Aaron, I'd say the first question is, can you explain as best you can, because you're deep in the forest for the trees, to somebody who doesn't know what we mean by agentic marketing? Remember, let's go all the way back up to the right altitude. People know what it's like to chat with ChatGPT or Claude or Gemini.
Even on a semi-advanced level, they prompt, they might attach a document, they might ask it to do some kind of tool call, whatever. What do you mean, and what's the difference between what people would normally experience with a chat bot version of AI versus an agentic ⁓ marketing agent?
Aaron (05:08)
Yeah, okay. ⁓ I'll do my best on kind of simplifying it. I think of this as, as there's two phases here. The first phase, I'll oversimplify and say marketing agents are going to do things that humans used to do, but way faster and likely way better too. But limited thereby humans could have done this by the way.
Okay, the thing it's going to go do for you, you could have done. Now, maybe you wouldn't have done it because it would have taken too long. You didn't have the time to do it, but you could have or your team could have. But yeah. Yeah.
Forum3 (05:39)
And Aaron, on that first phase when you say humans
could have done it, give me just one example of what you're talking about.
Aaron (05:47)
Yeah, sure. So like to make it super tangible for marketing, let's say you want to go run a marketing campaign, a marketing program, right? I have a thing I want to go do. I have an audience that I want to communicate a message to, right? Classic, like marketing teams do this all day long. You start with, I have a goal in mind, right? Like I want to get this message to this audience to try to achieve this business objective, whether it's to come buy my product or gain awareness or
Differentiate against competitor or whatever it is, right? So you're typically right You'll start with a plan like a brief right like the CMO says go get market penetration here or you know change the Perception of our product in this market whatever it is As you have a goal, you'll write this brief like a campaign brief, right? This is what we're trying to do Let's put a finer point on the audience. How are we gonna do it? What channels are we going to use etc? and And then what do you do where you're gonna go execute it? Alright, so that means
writing content, maybe buying media, know, whatever, like depending on maybe you're going to have to go do some sponsorship to conferences or you're going to run ads on TV or whatever. There's the actual execution, right? So that's that there's a plan and then an execution to oversimplify it. Today, humans do that like pre agents, pre AI humans wrote the plan. Humans did all of the execution. They were aided by software.
obviously in helping to do the execution. You're going to go buy ads somewhere. You're going have some software help you do that. You're going to do a podcast. Software is going help you do that. But humans are going to do it. Without a human, gets done. Agents, step one, the phase one of agents. Agents can do everything we just said. So CMO could just say, want more. The one paragraph of CMO might give a team that might lead to
500 hours of work over a six month period right that one paragraph that the CMO says in a meeting and says go do this The CMO can give that to an agent it can create the plan it can go execute all the steps of the plan without without humans if you want humans can be in the loop to etc, but Mostly the point being the agent does the thing that humans used to do ⁓ That's phase one exactly one thing that's worth noting like
Forum3 (07:43)
So that's phase one.
Aaron (07:48)
And we all experience this with ChatGPT. ChatGPT does things, like the classic consumer use case of ChatGPT is research. Go do a bunch of research for me and just give me the distilled answer. This is something we see all the time. My back's bothering me and this and that and whatever, and it's like, boom, here's the answer. That person could have done the research. It might have taken two hours, but they could have done it.
The reality is they wouldn't have, right? We're all using ChatGPT to now do things we wouldn't have done, but we could have done. The reason why we wouldn't have is we wouldn't have had the time to go do all the effort that AI can do for you, right? So this phase one is that it does things that humans could do, but because the human doesn't have to do it and I can run many agents all at the same time, I now get to do things that I wouldn't have done. So like my to-do list, right?
or organizations backlog of things they want to go do, you now get to go below the line. You get to go run experiments that you wouldn't have run before because your team is finite in size and you have to draw a line on what you're going to get done. I think it's worth pointing out that I use the words a person could do it because it matters in what's possible.
Okay, because phase two is doing things that weren't possible before. But these things were possible, but you weren't going to go do all the things. And that's the beauty of agents is that you're, you maybe could have gotten 10 things done this quarter with humans only. Now you get to do 50 things or a hundred things with agents. And so that does open up a lot of capabilities. So this phase one is still like massive in the transformation, but the things that get done are things that humans could have done. Right.
we're going to unpack this because there's a lot here. But one of the constructs there is that agents are doing things using the same tools that humans used to get the work done. Okay. And we see this with like ⁓ OpenClaw as an example. What's it doing? It's using the same tools you do. Like if you give it rights to your email, it goes into the browser and it sends emails on your behalf, right? It's using the same tools humans use to get things done.
In phase two, we now are using agents to do things that humans couldn't do. And that's because the agents are now part of the tooling. They're not sitting on top of the tooling, right? So one example would be in the world of marketing.
And is being able to, let's say you, in B2B, I'll use a B2B example here. You sponsor a conference, And so people are scanning badges, they're coming by your booth, right? At the end of the conference, you add a thousand people come by your booth. All right, in the what humans can do world, you wanna follow up with those thousand, right? Because they came by your booth, had a conversation with somebody.
Forum3 (10:15)
Thank
Aaron (10:29)
You're going to do something simple like, let's take the cream off the top, the executives that came by, I'm going to have a human hand write notes to them. I'm going to go and look at like, if I had my ideal customers profile come by the booth, I'm going to hand write notes to them. But of the thousand, maybe 800 don't fit into those categories. I'm going to send a pseudo personalized email to those 800. It says like, thanks for coming by the booth. If you'd like a demo.
Here's a link to the demo. Maybe it personalizes by industry or something like that. In phase two, AI can handwrite literally the perfect note to every one of those thousand attendees doing hours of research about them that you just, that you never, never like, you don't have enough people to go do that, right? And can do it in real time too. So maybe after scanning the badge,
it waits 30 minutes and then it sends that personalized note. And so your tooling itself is agentic instead of you're having an agent sit on top of your tooling. In the consumer world, might be like, yeah, sorry, go ahead. Yep.
Forum3 (11:27)
Aaron, can I interrupt for a moment? So
you just laid out two phases. And that second example you used with the 1,000, I mean, you're basically talking about agents being involved in the tooling, and you gave 1,000 customers coming to the conference. I mean, one could have a bunch of interns.
go do research on the thousand people and handwrite personalized messages. I thought what you were going to say was that there's a level of marketing that goes on with energetic marketing phase two, where it's truly humans could not do it. And maybe that's what you meant with that example. But like, I'm actually kind of want clarification on that.
Aaron (12:08)
Sure, sure. That's true. Maybe that wasn't the best. In the consumer world, we've talked for 30 years about right message, right moment, right time. This is like even maybe since the dawn of digital marketing, that's been a promise, probably more of an aspiration. In the consumer world, if you could have your
your product marketer sitting, you know, ready to talk to everybody who comes to your website and have an infinitely long conversation with them. Someone who is like, knows your product, the market, everything inside and out. And could have a one-on-one conversation with that consumer for the rest of that consumer's life with all of the context of all the conversations they've ever had with them. Right. That's what an agent can do. Right.
maybe I could hire 10 million product marketers, I could train all of them perfectly. But the scale piece, I guess, is maybe the tipping point here of like, that really true one to one communication with an infinite amount of context, right? Like the humans just don't have.
Forum3 (12:57)
Yeah, yeah, no, it's fair.
So it's scale and speed is what you're talking about in that instance, right? The speed with which it can happen and the scale and the personalization. And there's no question that the promise of AI in marketing has scalable personalization is one of them.
There are two areas of this where I see AI playing out in marketing. One is marketing strategy. And the other is the whole, you know, move 37, right? Like being able to analyze some other alternate, either go to I guess it's marketing strategy, some other marketing move that is more cost effective that you would never have thought of that a totally different market that you're
even going after that kind of AI insight into marketing.
Aaron (14:01)
Yeah, I think so. think the the scale and speed piece around testing, when I think about exploring those moves that you wouldn't have made, a big part of discovering a potential move is to do testing to see like, does this resonate or is this a market that's interesting for me? And again, having your agents enable you to do
type of scale of testing that you just you couldn't have done before that wouldn't have let you explore those you know really unlikely nooks and crannies way way down the list right and so I do think you need it deeply embedded in the tooling you know to kind of have it be able to explore those options and you need this like planning step of
an agent to go come up with crazy ideas that maybe even a human wouldn't have, and then go test those ideas. Right. And then come back to you and say, like, here's something that is a narrative violation and that you should go look at that might be something that's a really big opportunity for you.
Adam Brotman (14:57)
So Aaron, can you go back though, and that all made sense. What an agent can do that allows you to go faster and stronger, especially in phase two, but even in phase one to do things you wouldn't have done. And so they become like this co-worker slash augmentation of your entire staff, not just of an individual person. And it basically uplifts your.
Capability set your effectiveness as a marketing organization or as a company. So I get that. But go back again to somebody and explain to our audience, and explain to me even, what in your mind makes an agent an agent versus just a really capable frontier model chap?
Aaron (15:39)
Yeah, think you mean kind of comparing the call and response experience of a typical.
Adam Brotman (15:44)
No, I know that you're, sorry
Erin, you're too smart and you know all this stuff. Pretend you're talking to someone who doesn't know all these terms, like embedded in the tooling. And so, like when I'm getting at it, like, let me make it really simple for you to answer. Like, I'm on the ChatGPT Pro account. Is that an agent? Do I have an agent at my disposal?
Forum3 (15:52)
You're talking to Adam's mother.
Aaron (16:05)
Yeah, I mean, in my view, do like chat, ChatGPT in and of itself is an agent. there's one step here that really starts to unlock things that consumers can do in their ChatGPT experience as well as enterprises are doing actively is you can give it access to tools.
In a consumer simple example, I'm overwhelmed with my inbox. And I just got back from vacation and I want to get to inbox zero. You can connect chat GPT, your consumer version chat GPT to your Gmail. And you could say, first of all, let's do some smart stuff. So go look at all my past responses, you know, the last two months of past responses.
to see what do I seem to care about, what do I respond to? And I'll go to my inbox, my unread, and go ahead and like archive the things that I probably won't care about, and the ones that I am gonna care about, draft a response, put it in my drafts, and then I'll go review the drafts, and then maybe I'll be able to send them all. Or if you want to YOLO, could be like, go respond to all of them for me, right? And so that, ChatGPT can do the thing I just said.
And this goes back to doing things humans do today or candidate. You give it access to your Gmail tool that then lets it act as an agent for you on top of Gmail. Even the research case in my mind is it's an agent. If I say I want to go to Italy next summer and we like towns that look like Florence but it's too big and whatever, blah, blah.
and we're going during this week and I want to rent a car and plan the trip for me. mean, it's like, ⁓ it's going to do a bunch of internet research and like look at travel sites and stuff. And so it's an agent. mean, it's doing things that I would have had to go do as a human and it's doing it way faster. Yeah.
Adam Brotman (17:42)
Yeah.
Forum3 (17:52)
Bye. ⁓
Let me interrupt. Adam, do you agree with that definition?
Adam Brotman (17:59)
Yeah, I do. but I also want to call out the confusion that I think a lot of us have. I mean, us being those of us that think about and study AI and those of us that don't. Like, I agree with Aaron just said, I want to kind of point out why I agree with it. What Aaron is basically saying is that
The concept of what is agentic or what is an agent is a spectrum. But it definitely, almost by definition, has to have some kind of tool use. Even if it's like, how do I put this? Even if it's just going and doing some Aaron in the tool that you gave an example of, like, go do some research for me, the tool that it's using is internet search.
Right. So the fact that it has its own ability to do internet search and not just answer a question from its training data. Right. Or from or an attachment I gave it. So it can go outside of its current context window. Right. And go pull something into its context window and do something for me. In the one case, it's going into search and then compiling that, thinking it through and giving me an answer about a travel trip. And the other example you gave, which I was a good one, is I'm going to give it access to my inbox.
sent and inbox both, and I'm going to ask it to do something for me that it's, I'm giving it now in both cases, I gave it a task, not just a question, and it's completing that task for me with some tool use. I do feel that that is a form of agentic behavior on the part of the AI system. I went from question and answer to task, but as the task becomes
more complex with more tools and longer running and starts to involve things like memory, proactivity, scheduling. Now you're getting into like, it starts to feel like a co-worker. and I think Aaron, what you said.
Forum3 (19:43)
So that's actually,
that spectrum's really good. I'm glad you nailed that. That was well done, Adam.
Adam Brotman (19:49)
Yeah, as you know, we've been thinking a lot about it in the last couple of months.
Forum3 (19:52)
No, but I gotta say you even
clarified it for me because I heard Aaron's response and I was like, huh, do I really think of ChatGPT as an agent? I kind of don't like I, I consider deep research in agent, within ChatGPT, but your point of the Q and a distinguished from the task and then multitask with reasoning like that spectrum's a
Adam Brotman (20:04)
No.
Forum3 (20:17)
really nice spectrum.
Adam Brotman (20:18)
Yeah, and I think, so let's talk about that. Aaron, I'll kind of turn it back to you, since you are a esteemed guest here and you're an expert in this area. So let's go back to marketing, agentic marketing systems and the future of work as marketers. Let's flash forward to, I don't know what your timeline is for phase one or phase two and what becomes, but don't worry about that as much as the following. It's 2028.
Okay, I want you to project out to 2028. And it's not because of the Centrini research report, but because I just want to pick a timeframe that's about a year and a half to two years out and be like, I'm a marketer, I'm a CMO, or I'm just like a leader in a marketing department, maybe not the CMO. How is my life different in 2028 than it is in 2026? Or 2025? Because I think 2026 starts to take on some of the flavors of this stuff.
what does it look like? What does all this stuff we're talking about mean to me as a marketer?
Aaron (21:18)
Yeah, think that, so I'll answer this. I I liked your spectrum thing and I've got a couple of things on that too, but I think, so to answer this one, think, and I really do believe this. for those of us that are like deep in tech and building AI, I think the moment ChatGPT like came out, we all saw this future that I'm about to describe.
But in some ways, was like for me, it was like when I went to college 25 years ago, I knew quantum computers were going to be a thing in my lifetime, but I didn't know if it was going to be in 2010 or 2070. And so that was my moment with ChatGPT. I know this is going to happen in my lifetime, but I don't know if it's going to be 2028 or like 2070.
But now I'm pretty sure it's going be 2028. you're going to be an orchestrator of tasks of agents. And you mentioned like a marketing leader. The cool thing here, this is true about a marketing leader, but it's true about the entire marketing team, including the individual contributors, which is for me, that's the big unlock.
when we thought about like, how do we talk about our product? For a long time, we talked about everybody's a CMO now, because I think that ICs now have a team of amazing experts and as many as they want to spin up that work for them. And it's super empowering. ⁓ so in 2020, Agents, a team of agents. Exactly. A team of agents. Exactly. Yeah, exactly.
Adam Brotman (22:34)
And Aaron, you're talking about agents. When you say a team with your product, it's with agents.
Forum3 (22:40)
You're not humans, not humans.
Aaron (22:44)
And so I'll even answer your question. Forget about the marketing leader because they've had this superpower for a while. They've been limited by like how many teams they have and how many people they can hire. But like they at least had people that would go do work for them and they could think at a higher level. Every marketing individual contributor now has that superpower. And so their day looks like checking in on projects, reviewing. Reviewing becomes a massive
skill set like advancement that we all have to get good at. Again, leaders have been doing this for a little while. ICs maybe not as much, reviewing other people's work. We could also, if you want, you know, the function that's the furthest along here is engineering because of code gen and everything. that function is like people are doing this today. I think marketers and other functions are a little further behind. But
It's the bookends of the process that become the most important. What are we actually trying to do, the planning phase, and then the verification phase. And so an individual contributor will have like 20 projects that agents are working on for them, and they'll be checking in on the, you know, probably if they've already started on the verification phase, or maybe they're kicking off some new projects, and they're iterating on that planning phase. But most of our time,
today is spent in the execution phase for most individual contributors. That actually goes to zero. And you collaborate with an agent to get the right plan. And then it goes and does work. And then you review the output to figure out if we're happy with it. And so I think that's the, that change in process, 2028, every marketer is doing this where,
They're planning with agents and then they're reviewing output and figuring out like, did we hit kind of what we want?
Forum3 (24:26)
Aaron, can you talk seeing as we're talking about 2028? What does how does the marketing organization change? Like how does it change structurally? Count employee? Like how does it change?
Aaron (24:40)
Yeah, honestly, this is a real tough one the thing I just said I'm highly confident on, this one I'm like, I have a lot more uncertainty.
Forum3 (24:48)
You're gonna whatever you say
is wrong, but you're more informed about it this than both Adam and I and you're more informed about this than our audience. So give us your best guess. How does the marketing I mean, Adam and I have gotten very interested in what the future of work and what the future of organizations look like. And in my preface it to say that I think organizational structure is going to change as a result of AI and agentic AI.
Aaron (24:53)
Exactly.
Forum3 (25:14)
you know, everyone talks about layoffs, but I actually don't even think it's about layoffs. actually think the departmental structures change. So that's what's underlying my question. you have a lens that's very marketing centric. So you can talk both about the marketing department, but also sales and how it, how it fits in the organization. that's what I'm asking.
Aaron (25:34)
So I think that it's not just about marketing, it's about the whole company. I think about the customer experience. I build marketing technology and I'm a really a product focused kind of CEO. And so for me, everything is about the customer experience. Now, not the customer experience you designed, but the actual experience the customer had.
or has, right? And so if we look at this through the customer experience lens for a minute, I think, and then we layer in agents, we'll then see how the organs are reacting. And I'll use B2B as an example. In the B2B world, if a company gets large enough, they have from the beginning of the customer experience to the end or the tail of long-term customers,
There's probably like 30 people that have siloed jobs that own little tiny slices of this customer experience. there's the website. Okay. So there's a web team that owns the website. There's a product marketing team that owns the general positioning of the website. Once you get to the website, maybe there's a chat there with an SDR that you can talk to. And that person, all they do is just
interact with chat, know, like this one stage of the funnel, right? And then you aren't ready to buy yet, so you download some white papers. Well, again, there's a team, a content team that just focuses on top of the funnel white papers, right? And then, okay, you raise your hand and you fill out a demo form. Now you talk to an inbound SDR that's going to qualify you. And then they hand you off to an account executive who helps to close the deal. And then they hand you off deals closed, onboarding procurement, actually, before the deals close, you talk to the procurement team.
And then the deal closes and then you talk to your onboarding ⁓ specialist to get onboarded. And then now you have a CSM that helps you in your first year. And when you file a ticket, you get a support agent, a person that's focused on support. Long-winded, lots of people. Is that what I wanted? Like, can you give me more people? I want more silos of people to talk to. No, of course not. As a customer, that was a crap experience.
So then why did the brand do it? Why would we do this for 30 years? Well, because humans had to do the work and there's no one human that can do all those things for me. It's literally a unicorn. I can't train someone to be an SCR, an AE, a procurement, a product marketer, a CSM. So we said, we need humans. Humans have to specialize so they can be good at something. So we're going to specialize all these humans through this funnel. It leads to a horrible experience.
When I file my support ticket, they don't know what the salesperson sold me. They don't know what the procurement person cared about in the legal department. They don't know what my CSM is helping me achieve in my next goal. The thing I'm obviously getting at here is that an agent can be your partner through the entire customer, the same one. They know everything we've ever talked about. They can do procurement. They can do the in-down SDR. They can be my AE. They can be my CSM.
Adam Brotman (28:10)
Thank
Aaron (28:28)
They can handle my support tickets. And every time they talk to me, they know everything about me. They know everything we've talked about and what my priorities are. They know my whole team, right? So, back to the org. Well, marketing owned a piece of this. Sales owned a piece. Procurement owned a piece. CSMs owned a piece. these functions are gone. I just have one agent that I talk to that knows me.
And that's like the best customer experience. Like back to the human versus agent thing, wouldn't it be awesome if you could hire a human who was the world's best product marketer, SDR, A.E., procurement, CSM, onboarding, everything, and I could hire one of those for every single customer and they just sat on the customer's doorstep. And when they open the door, they're like, I'm here. What do you need? know, or they waited for the phone to call. That's what we have now. Right. And so there's no marketing anymore.
Forum3 (29:13)
So.
Aaron (29:19)
There's just the customer experience agent.
Forum3 (29:20)
So connect that, okay, ⁓
so what you just outlined, you articulated that really well. What does that mean for the organization in 2028?
Aaron (29:27)
Right.
I think this is where I don't know how it all lands, but behind that agent, let's talk about like how this thing would actually work. Behind that agent, there's a series of like strategic objectives that humans need to give that agent about each of the phases of this customer experience. Right. This thing isn't just your best friend. It needs to also
help the company achieve its financial goals and business objectives and everything, in addition to being a great experience for the customer. maybe marketing has input about strategy on a part of that funnel into this agent and says, when you're talking to the customer, make sure that you really highlight our differentiation against the competitors, right? And this is probably true when they're top of the funnel as well as when they're an actual customer.
But that's a domain that marketing typically would own is like product positioning. And maybe the sales leader is like, great, we really need to improve win rates in this market segment. So what are we going to strategically do with our sales agent to go get that done? So maybe there's still functional owners of things because that's how the business looks at things. But that's the kind of input you would give to the agent.
You would you want you these are our our goals this quarter, which would include conversion rates by segment and winning against these competitors and things like that, right? So I think that's maybe where we head is like you're telling these agents what you want to have happen just like a human would they're coming back with like real-world boots in the ground feedback like
Look, we've been trying to position you against the competitor, but nobody believes it. Like we got to improve our product better because it's not, we're not actually winning there. And you're telling a story that's not real, you know, that's like what an SDR might tell you. And that's what an agent's going to tell you. Right.
Forum3 (31:10)
I know.
Adam, I'm curious, what do you think? mean, we're coming to the end of the our time with Aaron, but given his answer, you and I have started to talk about this, but I'm curious, did Aaron's answer make you think about the future of the marketing department or the future of organizations differently?
Adam Brotman (31:30)
Yeah, mean, what Aaron's saying, which makes sense to me, is that whether you're a functional leader, a functional independent contributor,
or not, you become more of an outcome manager as opposed to a task manager or job manager. You think much more in terms of outcomes as opposed to tasks and jobs. Tasks and jobs always have led to outcomes, but you had to be, as Aaron pointed out really eloquently, as humans, we aren't generally.
super intelligent across every domain and across every function, even within a function. we have to be good at our tasks and our jobs. whereas if the agents, if the AI can be good at the tasks and the jobs, and we as humans can be abstracted up to being more of an outcome manager. So we understand what the company is trying to achieve. We understand what the function is trying to achieve.
what I heard Aaron saying is it might even change the nature of what we call functions and the role of functions. But even if it doesn't, even within those functions, you become much more outcome managers and task and job managers And either way, you become a manager. Like management skill becomes a thing. Yeah, like that wasn't, that isn't, I mean, it is the case for managers theoretically right now, but it's not the case for people that aren't
Forum3 (32:39)
matter.
Adam Brotman (32:47)
senior managers and it becomes a skill you have to have. mean, Andy, you and I do this because our career paths have led us to be managers for a lot of our careers. we sit in.
meetings with other humans all day thinking about like what is the point of this meeting and what am I trying to achieve with my company what am I trying to achieve with this meeting so we're thinking in terms of outcome and output management anyways whereas that's going to be the case for everybody all the time if you've got these agents I do have as we come to the end I have one last question for Aaron then we can bring it home which is
Forum3 (33:12)
Yeah.
All
right, hold on, because you have one last question. I have one last question. let's let's let's let's prove because we only have time for one question. What's your question?
Adam Brotman (33:32)
OK, well my question is, how realistic and where are we in the time scale to these agents being able to do this? We just described a scenario. Oh, yeah, I got this team of agents, and they're doing this stuff. But in my experience today, when we talk about in the spectrum of agents, agents like OpenClaw and Manus and you know.
Claude, coworking, code, whatever you want to call the cutting edge version of the spectrum of agents, they have varying capabilities today, varying reliability, varying overall ability to do the things we just described. My question is, where are we in this reality of these agents being able to be coworkers? That was my question, Andy. Maybe you have a different better question.
Forum3 (34:18)
I'm gonna, it's a good question.
Aaron, give us a succinct answer and then I'm gonna get my question in.
Aaron (34:23)
Okay, I'll go. I'm not good at brevity. I'll do my best. technology, will basically say is there. We could split hairs on some use cases, but the fundamental technology of these agents is there. You show me a problem of an agent. I told it to do something and look what it did. You know, we see this on LinkedIn all the time. You show me one of those problems.
Forum3 (34:28)
That you're, yeah, you're fine.
Aaron (34:49)
I will show you a problem with context, not a problem with technology. if you give me an outcome you want, and this is back to the management thing, it's like, is it the team's fault or the leader's fault? It's the leader's fault for not giving clear enough right instructions to the agentic team, in my view. Now, there's probably some outliers, like I'm being a little bombastic in saying the tech's completely there. There are probably some things.
that it can't do. But for most of what we do day to day in business, it can do. If you're inventing new drugs and stuff like that, maybe you still want humans doing some of work. But run of the mill business, as long as it's done digitally, robots are different, then we're pretty much there. That's what I say. And you've got to have the right context.
Adam Brotman (35:32)
Okay,
right on.
Forum3 (35:34)
All right, my question,
Aaron, is for CMOs and to a lesser extent CEOs, what concrete advice do you have for them as it relates to agentic marketing and what they should be doing in the next 90 days?
Aaron (35:48)
Yeah, I think the biggest thing that I see even in my organization and definitely in our customers and everything is just, and I think is a really big focus area for most executive teams and boards is just adoption. Like how much adoption are we getting across our company? You know, like what percentage of people are transferring? I love to this, we,
the, we become outcome managers, not task managers, Adam, that was great. So, you know, if we take our company, what percentage of our people are managing outcomes versus tasks, right? Like last week, hours spent by the whole company, percentage managing tasks versus outcomes, right? Like that slider needs to go to a hundred percent outcomes across everybody, like all the way, the intern, you know, it doesn't matter. So,
If I think about that as our KPI, that's the thing we need to be. I think we all would say the slider is not far enough over in our organizations. And that's probably true for everybody. like, I think we're trying to push that top down or at least in most orgs they are. I think the, like, we all know this, right? This is not the way to get people to do things. We, have to want to do it. And so I think the, the, the creativity here is
How do you get your company, every individual to want to be an outcome manager? And yeah.
Forum3 (37:04)
And so do you have
a tip or something they should go read or some meeting they should hold? give them one concrete tip, Aaron.
Aaron (37:13)
Sure. Yeah, think like, well, here's something I'm doing this month. I'm blocking off a week for the entire R &D organization. No one has any work assigned to them for that week. And so this is like 70 % developers, 30 % PM and test, something like that. We are going to get into groups, okay, like small teams of like three. There's going to be someone in that team that's done some level of
like agentic work, so has some experience with it. it's like a startup weekend. The company is going to pitch a bunch of ideas. Each of the teams is going to pick an idea, but here's the crux, okay? Back to the outcome thing. You can't manage any tasks. You have to get this thing done by only managing outcomes. Specifically what that means is vibe coding. you have to go to Claude code. You can't edit one line of code. You can't write one line of code.
You have like if you want something changed, you have to tell us to do it, right? So the you have to go get Claude code to generate all of the code, all of the tests, all of the designs. So the other thing is you can't write design documents. You have to just communicate. have to chat with Claude code, get all the designs, everything. Boom, boom, boom. And then what we're to do is like day one is to be like, OK, guys, day one, we're going to create our specs and our designs. So at the end of the day, I'm to go to each team and say, show me, show me your specs and designs. Show me the conversation you had with Claude.
Forum3 (38:15)
Yeah.
Aaron (38:29)
where you actually generated this. So it proved to me that you didn't actually write it. Tomorrow, we're going to write code. You're going to manage a team of coding agents to go implement these things. And then the last two days is going to be the verification phase of the test plan, automated tests, and everything it's called. So basically, it's a hands-on, but the leaders are participating and driving it. I think if people get a taste of this, they're going to love it.
And so you just need to like force that taste, right?
Forum3 (38:56)
Aaron, you want to give a short pitch for Pianza?
Aaron (39:01)
⁓ Inflection
Forum3 (39:02)
I'm sorry Inflection yes.
Aaron (39:03)
Yeah, yeah, yeah. Yeah, if you are, if you're B2B and you want to quickly get to this, this 2028 future we're talking about, then go to inflection.io. That's what we do.
Forum3 (39:16)
Great. Adam, we just had a great conversation with Aaron about agentic marketing, organizations, the future of work. What do you want to highlight for our what's jumped out at you that you want to?
Adam Brotman (39:26)
I would say I totally agree with Aaron's.
Forum3 (39:27)
highlight.
Adam Brotman (39:30)
fundamental premise and prediction that we have officially entered into an era where these AI tools are capable of doing work that used to only be able to be done by humans, actual work. that line, just to take the whole conversation we just had, that line of
Regardless of it can be better than humans or the same as humans like we've we've crossed into that line and so rather than there's a lot of like societal economic and Moral implications of that statement which way heavy I think on all of us and If it's true then the it changes the future of work pretty in the near future. It means that you have to start rethinking
about yourself as an output and outcome manager and as a manager period and in a way that you probably never had before unless you were a manager. And if you were a manager, you need to think about what you expect of your company and of your functions in a way that's different than before. That paradigm has shifted and it's coming at us really fast. I thought Aaron did a great job of articulating sort of
the realities of what that means and even thinking about it in phases. And I agree at the very end with Aaron's recommendation, which is like one piece of advice is you've got to use this stuff. just for a moment, block out the feelings you might have about that you're behind or what does this mean for you and your job? Like get.
your hands on this and start using it. And that gives you the best opportunity to be in the position of being.
you know, part of that future of work where like you're managing these systems. Cause if you're not using it and doing the kind of things Aaron just talked about on his sort of startup day hackathon idea, then you're just going to be like a deer in headlights and the rest of your coworkers and people are going to be like using these systems and you're going to feel left behind. And there's no reason for that. Like this is not hard to use technology. It uses simple
you know, plain spoken words in English and prompts and these agentic systems, they work the same way. Like we're talking about them as agents, not chatbots, but they work the same way you, you get yourself in front of an agent and you tell the agent what you want it to do. And you should, and Aaron said it well, like you got to give it context. Like a good prompt to a chatbot is give it context, give it
example, explain what you're trying to accomplish. Don't just treat it like a Google search query. Same thing with working with an agent, but you learn that by working with an agent. So I really just, that was my biggest takeaway from the conversation today.
but Aaron, I don't know if you agree with that.
Aaron (42:14)
I totally agree with it. And I think that the cool thing about where we are now, so we all started this journey, you know, when ChatGPT came out or even a little before that maybe, the cool thing about working with agents now is like all the stuff you just talked about, Adam, you you got to give a good context, you got to tell it what you want. There's nuances. You can just ask it. Like, I don't know what I need to tell you.
So tell me the types of things you need to know for me to go get this objective, right? Like it's this meta thing of like agents can help you create the context, right? And so ⁓ it's a, which is really cool. And at one point it didn't do that. And you had to like kind of have some kind of like, you know, this like homegrown skills here, but you can just say like, here's my outcome. Like what are the things I need to tell you to make sure that I get that outcome the way I want it to happen, you know, and it'll tell you.
Adam Brotman (42:44)
Yeah.
Yeah.
Yeah, we,
we, that's right, Aaron. We say that all the time in our boot camps. say, if you don't know how to work with the AI, just ask it how I should work with it, but explain what you're trying to accomplish. All you have to do is I'm trying to accomplish this. How, what's the best way to work with you? And so you got to at least be able to get out the outcome you're trying to get, but if you can do that, it can really help
Aaron (43:11)
Exactly.
Adam Brotman (43:21)
you.
Forum3 (43:22)
I thought what Aaron had to say about the definition of agent and the way that that conversation broke down really from Q &A to task to actually doing work and then ultimately to managing outcomes. I thought that was something that I wanted to highlight. I also thought that the conversation about
customer experience and how agents are going to affect organizational structure was something I've always been super interested in that. And I think that that's something that I want to highlight for the audience. That was an excellent conversation.
Adam Brotman (43:56)
for everyone in our audience, if you want to learn more about what Aaron's doing, go to inflection.io. If you want to learn more about what Forum 3 is doing, go to forum3.com and you can join our community, you can join our newsletter and thank you for joining us on this episode, Aaron and everyone who listened.
Aaron (44:16)
Awesome, thanks so much, Adam. This was super fun.
Adam Brotman (44:18)
Great.
Forum3 (44:18)
thank you all to listening.
to AI First with Adam and Andy. For more resources on how to become AI First, you can visit our website, form3.com, download case studies, research briefing and executive summaries, and join our email list. Of course, we invite you to connect with our AI First community, a curated hub and network for leaders turning AI hype into action. We truly believe you can't over invest in AI learning.