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.
Andy Sack (00:00)
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 co-host, 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.
Good morning, Adam.
Adam Brotman (00:32)
Good morning, good afternoon.
Andy Sack (00:33)
Whichever the case may be, wherever you are in the world.
Adam Brotman (00:34)
Glad to be back. Yeah, wherever
you are. Yeah, we've recorded a... We have one episode we recorded in the New Year, which I'm excited to come out after this or before this, but I'm glad to be doing another mini episode with you.
Andy Sack (00:47)
Yeah, and so we talked this morning, my time. We're recording this podcast. I happen to be in New Zealand and happen to be growing a beard. Those things are unrelated, but just thought I'd share that with the audience. And we talked this morning about really the progress at the end of the last year of the models, particularly in the domain of coding, and really
you playing over the holidays and I just thought you should share what you built over the holidays and some of the reflections you had. The broad topic that we want to just cover pretty briefly is agents for executives, business executives is the broad topic. So Adam, why don't you talk about your holiday and what you built?
Adam Brotman (01:35)
Sure, let's deconstruct this a little bit to say,
Remember a year ago, we were all about the year of the agents. And there's this sort of, not fantasy, because I think it's becoming more and more of reality, but a little bit of an aspirational notion of like, there'll be these agents, and they'll be like almost like digital employees, and they'll come and they'll help you do your job better, or they might just do a job instead of you having to hire somebody or whatever. So this concept of agents is out there.
Andy Sack (02:03)
that concept of teammate or digital employee is something that we're excited about at Form 3.
Adam Brotman (02:09)
We are. It goes all the way back to our book where we talk to Reid Hoffman. And he's like, in a couple of years, everyone's going to have this. There's going be agents everywhere. And everyone's going to have a special agent to help them 10x their work. And so we've been on the scent of this agent thing. But it's confusing because, OK, so are there agents in the world right now? Is a chatbot an agent? What is an agent? It's confusing, I think, to the average person.
that's not living in the bubble we're living in. And it's even confusing for us. So let's talk about this for a second. The thing where there are agents, and this came up over the holidays, to get back to your point, is coding agents. If you're a software programmer, coding agents are the furthest along of any kind of agent right now. And it was all over all of our social media feeds about Claude Opus 4.5 because it is an amazing coding agent. If you use Claude code,
with Opus 4.5, people were like, holy crap, I can give it a coding assignment and it will code for hours. And I don't have to write any more code anymore. And by the way, you and I are not software programmers, so we don't play with Claude code or cursor. But we do play with things like lovable and Replet, which I'll get back to you. So we've even noticed, not just from our social media feeds, but ourselves, that coding agents where you just
describe in plain English, not code, plain English what you want this coding agent to go build with software code, and it just goes and does it. And it does it increasingly well on the first try, and debug it and make iterations increasingly easy. the notion of an agent when it comes to coding is alive and well. Like that's not even like a debate, it's alive and well. And...
But the notion of an, for us non-technical folks, so business leaders, if you're a CEO, CMO, vice president of marketing, vice president of HR, like sales leader, whatever, what does a coding agent have to do with you? And we've been talking about how, well, coding agents are instructive because while they're the furthest along, if you know a little bit about how to use AI, if you're a little bit workflow proficient,
in AI, which we'll give some examples in a second. Like you can emulate a coding agent, not perfectly. It's not like, you know, describe something, walk away for a few hours, come back and it's done. But you can come close if you know what you're doing. And so that's a really interesting discussion, which is there's no cursor or lovable yet for non-coding agentic tasks.
I think everyone's sort of working on it. I think there's some. the legal profession, there's a thing called Harvey. And you're starting to see these sort of sophisticated wrapper apps kind of act like agents in non-coding situations more and more, where you sort of give it a task or a thing you want to accomplish, and it goes off and does it and comes back. they're not really, other than a few exceptions, they're not really existing in the wild.
You and I should talk a little bit about how we kind of had the aha moment of like, if you kind of know how to prompt and you kind of understand what these models are capable of and you can sort of do these workflows, you can sort of emulate a coding agent.
Andy Sack (05:17)
Yeah, ⁓ I mean, I think that conversation that we've had, it's hard to do this on a podcast, but I'll attempt, which is on one side, you've got hard coding in workflows to a job to be done for a specific industry. in our case, we like consumer facing businesses, restaurant and retail. So you could imagine.
sort of labor scheduling, right? So you could actually hard code a labor scheduling app. On the other hand, there's just, I'll call it, really smart, agentic work that an executive can do with context. we think that that area, like there's going to be a blending of those things, that that's sort of a spectrum.
The workflow pieces require actual engineers today. You can actually do a lot with the models as is if you know how to use them and how to set context. And we at Forum 3 are experimenting with that.
Adam Brotman (06:12)
That's right. we spent a lot of time in 2024, and particularly 2025, advising consumer brands on teaching them how to fish, saying, oh, let's help you think about how you get your whole company
to maximize the opportunity for making better decisions faster and being more productive and all the good things that come if you get a whole company or a whole team proficient on AI. And a lot of the things we're trying to get them to do, we're realizing, in a way, we're trying to get them to be smart enough about the landscape and what's possible so they can become their own Claude code of business.
It didn't occur to us until this holidays when we were like, because we were talking to a lot of brands in the last couple of months about, what would you like to get out of AI? And just curious why they weren't doing more with it. And they would describe, I wish I could use it to help me optimize my store portfolio or to do segmentation of my customers for personalized marketing or to do labor scheduling better or to do whatever. And we were like, you can do that.
And like, here, let me show you. But we realized as we were showing it to them, it was becoming advanced for them. They're like, ⁓ it's not like needed a, you didn't even need necessarily a developer. But it was like, for example, let's talk in real terms here. one of the examples was this weekend, I wanted to see if I could create an app that would help me put in any restaurant brand, any. It doesn't matter, public, private, local, national, international.
any restaurant brand and create an app where it would analyze the strengths and weaknesses of that, how that brand is performing in general against a benchmark of the 19 top performing public companies. I was like, that's interesting. And by the way, I just wanted it for myself. I was curious to if I could build it. And I went through a workflow and I built it with Lovable. I vibe coded it in Lovable.
But to get there, I had to understand five or six pieces of good prompt craft and AI workflow. Like I was using Notebook LM to like take the transcripts of the 19 top companies and then use that to create a formula and then convert that into a meta prompt that I then put into Lovable. And then Lovable took me all the way down the field, but did still say, you need to go get a perplexity API key and
And then I was like, but it was amazing. I went through all these steps that to me were at the edge of what I'm capable of. So I know they're beyond what most, I'm not that advanced. I'm not an engineer, so I'm not trying to pat myself on the back. But I'm saying when it comes to complex workflow, understanding the landscape, understanding the difference between what a reasoning model can do, the benefits of a large context window in Notebook, LM, et cetera, I go on and on. I'm at the, you know,
1 % level of non-technical people just to kind of understand what's possible. And there's a lot of people that are better than me, but I'm not a dev. But I was able to create this app in about 45 minutes this weekend. And I shared it with you and our team, and everyone was like, kind of raising eyebrows. So it wasn't an agent. I mean, the lovable part was a coding agent to build this app. But the raw materials to like, if I was baking something or making a meal, like,
Andy Sack (09:15)
Yep.
Adam Brotman (09:29)
The raw materials, like I had to know which ingredients to put together and where to find them and how to put them together. But at the end, it was like the lovable robot that like was an agent. But the whole thing wasn't an agent. And I think that that's the big aha, which is like, while we're waiting for agents to come, in forum three, we might even build some agents for some folks and whatever. It's your ability to know, to just use this stuff and get proficient.
is kind of the Claude code of business. And that was a big aha for us. It doesn't mean that people can't do it themselves, like we've been trying to teach them. But it also probably means they need a little bit of help right now while we're waiting for the Claude code of agents. And it's a bit of a mouthful, but does that make sense, what I'm saying?
Andy Sack (10:11)
Yeah.
It does. It makes sense to me. I'm not sure it'll make sense to our audience, but we're going to put this episode out there and hope that it does. I mean, building on what you're saying is the image that I had was the Swanson's TV dinners, which is basically what people are what we're busy talking to executives and we're trying to coach them about what's possible and
Adam Brotman (10:20)
Yeah. you
Andy Sack (10:36)
coach them about how to do certain things, how to work differently with AI by prompt engineering and context engineering. And in many ways, what we're talking about is many of our customers, just want the TV dinner, like just give it to me. And that makes sense.
Adam Brotman (10:51)
Yeah. they're happy.
Yes, they want the TV dinner, or maybe even also they might just want us to come in their kitchen and cook the non-TV dinner for them. And we're like, but you can do it. This isn't that hard. they're like, it's harder than you think. you get, by the way, you end up.
Andy Sack (11:09)
Would you just do
it for me?
Adam Brotman (11:11)
Yeah, and also
some of the stuff that we encounter, we've talked about this in prior episodes, is there's this notion of not only do we sometimes, not always, have a better understanding than our clients of how to pull all these things together, but our business understanding, because we've been in the seats of our clients and been the chief digital officer, et cetera, we...
We also sort of know what ingredients you need. And for example, a big ingredient everybody thinks they need is, I need to have all my data in one place all cleaned up to do anything. And the truth is, no, you don't. That would be nice. And eventually you might. Particularly if you do want a true hands off the wheel agent to just go away and do everything automatically, yeah, sure. If it's going to be drawn on data, it should be clean data. It needs to be accessible.
In the meantime, your last 90 days of data can take you, talk about 80-20 rule, it may even be 90-10 rule. It can take you a long way down the field. it's just this episode, think, is dedicated to the notion of while everybody's hearing about these vibe coding tools and Claude code and all this stuff, and they think that it's just applicable to coding agents, you and I are.
Andy Sack (12:03)
Yeah, yeah.
Adam Brotman (12:25)
And we were inspired, and I want to give a shout out to Paul Raitzer, because we listened to his podcast, and he's an upcoming guest on our show. And he talks about the same thing. He talked about, like, over the holidays on one of his recent episodes how he has this CEO custom GPT, and he was using it with reasoning and thinking models to do all this work that would have saved him months of time over the holidays while he was not even
bothering anybody on the scene, was just doing it himself. And he was remarking about how it was kind of the Claude code of business workflow. So I think our biggest point to people is that we're not there with agents yet, but when you hear about all these advances in coding agents, it should inspire you to be like, I wonder if I could do that, or I wonder if could have somebody do that for me. Because these AI systems are
a lot more capable than you realize, but you do have to sort of know how to be a cook.
Andy Sack (13:24)
That's a great spot to end.
Thanks for listening to AI First with Adam and Andy. For more resources on how to become AI First, you can visit our website, www.form3.com, download case studies, research briefing, executive summaries, and join our email list. We also invite you to connect with our AI First community.
Curated hub and network for leaders turning AI hype into action. We truly believe you can't over invest in AI learning.