From CRM to AI Workforce — Fireside Chat with Kipp Bodnar & Tooba Durraze

Tooba Durraze:
Thanks, Julia. Thanks for having us. I would say my tool is my own tool right now, just because we geek out of using it.

Kipp Bodnar:
Self-serving? Sounds awesome.

Tooba Durraze:
I think, yes, early-stage founders, of course, will always say our tool is the tool that’s front of mind for us. Well, I’m very excited for this conversation today, Kip.
I have to say, the last time we spoke, I was… extremely excited about how pragmatic our conversation was, so I think this is going to be a treat for folks in the room, essentially.
By way of introduction, again, my name’s Chuba, I’m the founder and CEO of Amoeba AI, I sideline as a geek, I studied a lot of neurosymbolic AI and traditional AI ML techniques.
I’m excited to talk to Kip, who’s been the CMO of HubSpot, actually needs not a lot of introduction, but Kip, why don’t you say a couple of words about your background, and then we can kind of jump in.

Kipp Bodnar:
Hey, yeah, I’m Kip, I run marketing at HubSpot, host a podcast called Marketing Against the Grain, and I spend a lot of time doing marketing and AI stuff, and I’m happy to be here.

Tooba Durraze:
Great. Okay, so first question, and this is, you know, I’ve been kind of dying to re-ask you this. You’ve watched a ton of waves go by. And a ton of claims around how X is gonna kill Y, or this is gonna change marketing forever, there’s a new thing dead every other time we look on LinkedIn these days.
I am just curious, what makes this one different for you, and what makes you feel like this might not be hype in something different? Or you might feel like it’s hype, but I’m curious on your thoughts on that.

Kipp Bodnar:
You talk about, like, the AI wave overall, right?

Tooba Durraze:
I would say the AI wave overall, yes.

Kipp Bodnar:
Yeah, look, I mean, it’s just, it’s too big. You know, it’s akin to the… the foundation of the internet. I mean, I think it’s bigger than the mobile app revolution, which kind of was overhyped, even though it ended up being big. We just turned out the average person didn’t need that many mobile apps, and so it became a different thing.
For me, you know. AI is a massive way because it changes the physics of knowledge work. Right? You used to scale through hours and time and money, and you would have this, like, diminishing curve of, like, context and skill that would go down as you were trying to scale that out.
And now, you can really scale a volume of work at a high context level, because you can have… the people I’m talking to that are deeply skilled, they’re not, like, 1 or 2 times more productive, they’re 10, 100 times more productive.
It’s a massive change to how we work, and it’s just… it’s literally going to disrupt every aspect of how companies build products, sell products, market products. It’s I don’t think there’s any close that this one’s actually overhyped.
It’ll have its waves, and Many bubbles with inside of it, but it’s going to be the foundational theme of the next… next decade.

Tooba Durraze:
Yeah, totally. I’m glad you think it’s not hype, because that would set up.

Kipp Bodnar:
It would stick for you, right?

Tooba Durraze:
Yes.

Kipp Bodnar:
Totally.

Tooba Durraze:
We were having a conversation yesterday on a panel, and folks were talking about it. I think the earlier conversation here was also talking about this, like, agent-to-agent architecture, and that we’ll live in a world where, you know, we’ll all be at a beach drinking water from coconuts, and agents will be buying and selling to each other.
I’m of the belief that’s a little bit further away, although I feel like…

Kipp Bodnar:
A long ways away.

Tooba Durraze:
inherently headed there. You brought up something in our conversation, which is the technology may be ready, but the humans are not ready right now. I’m curious on what that looks like.
I’d be also curious for your lens inside HubSpot, where I’m sure you have a combination of folks who are really into the way the technology is progressing, as well as maybe some skeptics in terms of the capabilities of the technology. So two lenses. One is, kind of, how do you think, where do you think humans are in context of that technology?
And second, in your lens as a CMO, within your org itself, how do you kind of level set that on both sides of the spectrum?

Kipp Bodnar:
Yeah, a few answers to this. One, I talked with an awesome AI founder recently, and, you know, he had a simple statement that AI can do anything a human can do on a computer. And I largely agree with that. Right? And so if you believe that, then we are way behind the opportunity. You know, like, that’s not how the world works right now.
And even, like, Anthropic, a couple weeks ago, they published a study of, like, AI impact on different types of jobs and everything, and it was, like, actual impact versus perceived impact. And there was this huge delta, and that delta’s largely because people haven’t changed and people haven’t adopted, right? And change is hard.
And, you know, the laws of physics still apply to, like, change and humans and everything else, and so I find that it’s just very challenging That the technology could stop getting better today, and we have.

Tooba Durraze:
Yep.

Kipp Bodnar:
decade worth of work that we could do on the technology right now, and I think humans are a little overwhelmed. I think something that I believe is, like. I spend a lot of time in the Bay Area, but I also spend a lot of time in what I’d call, like, traditional rest of America, and the gap between those two areas have never been broader.
You know, I walk around the Bay, and it’s like, everybody looks real stressed. It looks like, like, there’s a pending doom happening at any moment, and the rest of the world’s, like, fairly happy, I think, and it’s just a very interesting dichotomy of what’s happening right now.

Tooba Durraze:
Yes, I’m from the Bay, and I blame coffee consumption on us looking stressed all the time. I’m curious on what, per capita, like, how much more coffee we consume, by the way, but, yeah, I agree with you. There’s a whole world out there where they’re just starting to… and I think the analogy you used at that time is.
Something about, kind of like, golfers, like, folks booking tea time for golfers is something you had brought up. Are they concerned about what the latest Opus model is?

Kipp Bodnar:
like, I think people who work in technology forget that the… all of the far-reaching, applications to that technology, right?
Like, if you’re going to a golf course, you’re going to a kid’s event center, like, they have to, like, figure out how to take your money, they gotta figure out how to, like, have a basic calendar, and this is not their primary job. Right? And so, it’s gonna take a long time for those things to transform and evolve. Right?
And it’s not just going… just because the technology exists doesn’t mean it happens. Like, email marketing has been around for, like, 40-some years, and it’s still a thing. And there are still people who don’t do a ton of email marketing, right? Like, this curve will be faster than that, but it’s gonna take time.

Tooba Durraze:
I know we have a lot of founders tuning in right now, as well as, obviously, folks who have kind of been in the industry for a while. What is one advice you would give to the founders that are building right now? Presumably, you might say building for the wrong audience. There’s a lot of pressure in the Bay Area from founders who are building.
to be, you know, building on the latest tech and constantly evolving that. I think I’ve heard if you’re not launching a new product every week, you’re behind already. What is the advice that you would give them? Because presumably there’s a whole world out there that needs to also kind of catch up.

Kipp Bodnar:
Right, it’s like, just because it’s easier to build and ship things, From a digital technology perspective. doesn’t mean it’s any easier to distribute those things or get people to use them. In fact, it’s harder, right? The competition’s higher, and so those things are much harder.
And so my advice would be, one, understand who your real customer is, and your customer isn’t other startups, or tech stars, or YC, or, like, most of the founders I talk to, and I talk to probably, like, five a week, it’s like, they’re very much in this bubble.

Tooba Durraze:
Yep.

Kipp Bodnar:
Of, like, oh, they’re on, like… you’re not, like, the .1%, you’re, like, the .000001% in how you’re thinking about things, and your job is to, like, at least get back to the 1%, so that you have some semblance of a market, and then you have to figure out Like, how are people gonna find your product? And how are they gonna adopt it?
And do you need humans to help them adopt it?

Tooba Durraze:
Yeah.

Kipp Bodnar:
The dirty secret of a lot of… most of the AI tooling adopt… adoption is that it’s very human-dependent.

Tooba Durraze:
Yes.

Kipp Bodnar:
It’s, oh, some technical person comes into a company, does a bunch of custom setup, and all of this work, and it’s like, it’s not that dissimilar from, like, the early days of cloud or the early days of on-premises technology.

Tooba Durraze:
Yeah, totally. I’m so happy you said that, because that is indeed the dirty truth behind a lot of AI, and while we’re working towards potentially having these systems be completely autonomous in a way where humans don’t have to, right now, anything that human touches needs a human touch when it comes to AI, I would say.

Kipp Bodnar:
And there’s a lot of hard parts to it. You gotta get all of your data integrated, people have to trust you enough to integrate some data, which most companies don’t. Most startups, like, have no trust, they have no distribution, and they have no adoption motion.
So it’s like, would you… should you be spending 100% of your time just shipping product, or should you be spending a lot of your time also figuring out those other things?

Tooba Durraze:
Totally. I completely agree with that.

Tooba Durraze:
Okay, let’s move on to maybe something that’s front of mind for a lot of folks, and we’ve talked… it’s front of mind, we’ve talked about it for a year, which is, like, the transition from AI being experimental to, like, actually you using it for things that you feel like are going to be providing value versus, like, just tinkering.
You talked about a couple of bets at HubSpot. One you talked about was AI search optimization, like, the gold standard of, like, trying to get to one-on-one personalized messaging. what… at what moment to you did they feel like they weren’t experimental anymore?
And the caveat here is that, we get a lot of inbound that is potentially wanting to be more personalized, but is not a lot, from a lot of different vendors, so I’m curious, in your experience of… at HubSpot, where do you feel like you made the shift from it being experimental to not?

Kipp Bodnar:
Yeah, so you move things from being experimental to not when they go from being very, very hard to not that hard to do, right? And so… when they’re experimental, you’re learning the foundational primitives of those things. So, like, you talk about, like, personalization in your marketing. Like, let’s be very specific.
Like, let’s talk about when to do email marketing in a personalized way. Seems reasonable. We’ve had, like, quasi-personalized email for a long time. Now you can do pretty good personalized email. What you need to do personalized email really well is you basically have to have all the underlying data about each individual person you’re emailing.
Yeah. And data that is unique and interesting, and that matters to them, not just some generic set of data, right?

Tooba Durraze:
Yeah.

Kipp Bodnar:
Based on all of that information, you need to merge that with a prompt. Yep.
Put each individual prompt and data set by each individual person through the Cloud API, or the OpenAI API, or whatever, generate a actual message, have humans review those messages to understand what’s good or bad, and do the eval process, and then you start sending them out.
And then after you’ve done that for, like, a while, it’s actually no longer an experiment. But that’s, like, pretty hard, still. And, you know, even if you’re a company building products to do some of this stuff, it still takes a little while. It’s not like instant implementation.

Tooba Durraze:
Yeah, again, one of the things you spoke about was AEO and GEO as well. They’re actually, weirdly enough, varying beliefs. Some people feel like it’s not really a thing, it’s more like stakeholders. Some people feel like, yeah, this is how you get ahead.
Where do you land on that spectrum, and what would you say is, like, the one thing someone in the room can do to show up better in, kind of, search results that you’re doing as well?

Kipp Bodnar:
Yeah, yeah, so there’s a few things. Maybe, like… last fall, I was… we were having conversations with a bunch of different folks who own different marketing agencies. They’re partners of ours, and they’re awesome people. And one of, like, one of them was, like, is a pretty big skeptic, long-time marketer, and he was just like, look.
AEO has been the most transformative thing to my business in the last 10 years.

Tooba Durraze:
Whoa.

Kipp Bodnar:
And… he’s like, I’ve got… my programmatic content plays, my visibility, I have, you know, 3X to my customers, and it was fundamentally because, one, it’s a thing, and two. it’s like a lesson of marketing. He focused time and attention on it versus, like, doing a little bit of a bunch of different things well.
And so, you’re asking, like, what the heck do you do on the AO side of things? You need to understand, first, like, your visibility. There’s, like, we have an AO grader product we’re gonna have soon, coming in the next couple weeks, an awesome, like, full-blown AO product, that I would highly recommend everybody checking out, but it’s like.
You have to go and understand where are the gaps by specific prompt, right? Like, if you look, the average Google search was, like, 4 to 6 words, the average search prompt is, like, 40 to 60 words, right? There’s an order of magnitude of difference.
And so, understand how you are not being showing up, or showing up, and we call that mentions and citations. Like, how are you being mentioned? And is the sentiment around that mention positive, negative, if you are… or negative if you are being mentioned? And then, are you being cited? And that’s like a link back.
And you essentially have to look On a prompt-by-prompt basis. What are they… how am I being referenced if I am being referenced? If I am or I am not, what are the source materials for that specific prompt that are being used, and how do I influence those source materials?
How do I make sure I have content on my own website that is chunked in plain English that a LLM would crawl and find to make myself a possible answer? And then how do I continue to have more of a consensus view outside of my website. So that might be Reddit, that might be affiliate marketing sites, could be a whole host of things.
I’m giving you, like, the 3-minute version here. Those are the basic things that you’re going to want to need to go and do.

Tooba Durraze:
Hey, that’s great. You basically gave us a playbook, essentially, that somebody can take and apply. I think the biggest thing…

Kipp Bodnar:
I gave you the Varia Bridge. There’s lots of information out there. There’s a bunch on our YouTube and website. Highly recommend checking it out.

Tooba Durraze:
Bigot says, you are a believer. It is not snake oil to you.

Kipp Bodnar:
Look, we’ve seen massive growth. What I guess I can say is it went from, like. low visibility, low traffic, to, like, material visibility and material traffic to us, and we’re, like, a larger organization, so I think that foots for almost anybody.

Tooba Durraze:
Yeah, so speaking of then all, like, we were bringing up LLMs and kind of all the investments that, even with HubSpot, you guys are making in terms of your own technology. Yeah. The… the answer to everything is obviously data, but folks have been talking a lot about this idea of, like, context engines on top of your data.
You spoke about that being a core bet at HubSpot, so I’m curious on your thoughts on that, and then I have a follow-up question to that.

Kipp Bodnar:
Yeah, look… data is a input to context. Context is not just data, right? You know, context is… it’s representative of, like. the actual first and third-party data you have, but it’s also representative of how people work, what you know about how people work, what works, benchmark data across different company sets. I mean, part of, you know.
the benefit of… I think you’re going to see a lot of incumbent companies lean on context, because it’s an inherent unfair advantage that incumbents have. Versus startups, right? Like, if you’re going back and forth, and if you’re a startup, your job is to try to get a data flywheel going quickly, so you can build… build context of your own.
And so then you’re like, okay, well, in theory, this sounds good, but, like. is it actually good?
And what I will tell you is, like, we have, you know, we have this AAO tool in early, early, like, private access, where we’ve got hundreds of people using it before we put it out, and it’s like… we recommend, based on everything we know about you, the prompts that you should actually care about and track for all your AEO.
Like, we have an absurdly high accept rate. like, 95 plus percent accept rate on the prompts we recommend, because we have really great context, right? Like, and that’s… and we know your business, and we know what your customers are telling you, the conversations you’re having with them, all of those things, right?
And I think that’s emblematic of, like, that’s just, like, one little part of, like, the work you do. where context makes a big difference across just your marketing, let alone sales conversations, customer support conversations, so I think context is… something that a lot of companies have.
A lot of it’s going to be about how well you can deploy it, and how valuable you can make it as part of, like, the agent experience.

Tooba Durraze:
Totally. And for all the folks who are building… because Amoeba sits kind of on top between your system of record and your system of intelligence, and a lot of folks are building agents on top. I will say, and we integrate with HubSpot, we pull a ton of data out of HubSpot.
There is a richness of metadata that HubSpot offers, that allows agents on top to operate fundamentally differently than some other platforms, so I think that from a data geek myself, your investment in data and, like, kind of how that’s progressed shows, which I think is foundational to how the new world will operate, so even from a…

Kipp Bodnar:
Closed… I think closed data sets are going to kind of die off, because you want to extend, and I think the thing that’s being missed in all of this is there’s just, like… there’s gonna be so many new markets, so many new use cases, like, we are going to do 100x more, you know?
Like, we’re not gonna do the same stuff better, and we’re not gonna do stuff 10% more, we are gonna do stuff 100x more, and that’s gonna unlock a bunch of opportunity.

Tooba Durraze:
Yeah, no, totally. I agree. So people should stop locking their data down, thinking that the data is the mood. It’s actually investment in that context and, like, the use of that context.

Kipp Bodnar:
It’s the value derived from the data, and I’m sure… and I believe, have faith in capitalism and humanity that everybody can figure out the right exchange of economics to make that work.

Tooba Durraze:
Yeah. No, totally. And also, yes, we used to say… people used to say DeHa’s like oil. I would always make the joke that has, like, crude oil, if you put it in your car, your car’s gonna explode if it’s not in the right shape, etc. So, even for, like, vendors in the room right now.
The shape matters, and companies that invest in the shape shows to, like, agents and other interaction models.

Kipp Bodnar:
Yes, I think that.

Tooba Durraze:
Let’s talk about, because you are such a practical person, One of the things I wanted to touch on, there are a lot of conversations around, you know, existing agents that are very widely in use, which are kind of your customer support agents, your normal sales prospecting agents, marketing agents.
you look at them as kind of, like, the lowest hanging fruit, or picking off kind of the most obvious use cases. What are the less obvious ones that, you know, people haven’t picked off on yet? And then, for the more obvious ones, I’m also curious on where do you think they have room to grow, or they’re breaking down right now?
I asked you a very huge question.

Kipp Bodnar:
Yeah, look, there’s a lot of parts to that question. One is, like. even the obvious agent use cases are, like, 1 tenth of 1% into their lifecycle and value, like, it is very, very early.
We… you look at something like, you know, we have a great customer support agent that tons of people use, has really… is, like, I think the highest resolution rate, and… just the amount of work that you need to do around email, and email permissioning, and all of the ways people want to interact with an agent via email, like, that’s a massive amount of work, and that’s just, like, one little tiny part of an agent, right?
And so, these, like, you know, kind of broader use case agents are… are big, long-term products, they’re not these, like, disposable things. I would also say, I talked… getting out of the Bay Area, most people still do not want, really, autonomous agents. Most of humanity has no interest in autonomous agents.
They do not trust them, they are not willing to let go of control.
So I think regardless of the use case you pick, you are going to see the companies that win are not going to win by having the best agent, they’re going to win by having the best human-agent integration, it’s gonna be the easiest to get the data into and context into, it’s gonna be the easiest to get started with.
Like, those are the things that are fundamentally, like, going to matter, almost even more so than the edge cases. That being said. there are endless edge cases.
Like, I’ve talked to companies who are doing… like, there’s one that’s doing, like, an IT network administrator agent to better… better process, you know, network performance and make that all more efficient. Like, the… they’re endless, you know?
And I think what you have to say is, what is a really important job that gets done that is, like, has some kind rules to it. Like, there’s real feedback loops, so you can potentially build a higher quality agent experience around it.

Tooba Durraze:
Yeah, I agree. I was trying to convince folks that, you know, offload things to agents yesterday in the panel, and I realized, okay, we’re not there yet. Psychology and, like, kind of technology-wise.

Kipp Bodnar:
I would really encourage people building at the forefront of AI to go and spend some time with humans.

Tooba Durraze:
Yeah, no, I… we… and again, our products, generally, to your point, knowing the job to be done is, like, the most important piece, and then building kind of around that, and who is doing the job is probably the best suited to understand what the agents need to orchestrate.
Okay, let’s switch topics to you as a leader a little bit, and we talked about, kind of, how hiring practices are changing a lot as well with these agents. One of the things that you spoke about is, well, you yourself were talking about all your coding projects, etc. everyone is… building is democratized now.
You spoke about something very interesting, which is, of all the candidates that kind of reach out to you, maybe only one in 100 kind of sends you something when you ask to see kind of what their work is. Can you speak a little bit about that, and kind of how that trend is evolving or shifting for leaders?

Kipp Bodnar:
Yeah, so I think what’s really evolving in the world is that you know, disposable code is, like, one of the biggest changes in the world, as I look at it, is that you can just, you know, write code at a much lower cost, and so forth. It can be disposable, just like, like, a paper cup. And… What that means is that you can show and not have to tell.
And you can demonstrate your quality of thought, your understanding of the problem through a awesome deck, a really great prototype. There’s a bunch of different ways, right? I would prefer the prototype, or agent, or what have you.
So, like, you know, I bet you 100 to 200 people a month DM me on, like, on LinkedIn around different jobs and everything, and I always have the same response to them. It’s like, send me something you build. It’s literally, that’s That’s all it is. And most people don’t reply. To me, once I… it’s like, send me something you’ve built with AI.

Tooba Durraze:
Which is shocking to me, given the job market, that people don’t.

Kipp Bodnar:
Well, it’s like, it’s not an easy thing you’re asking people to do. But, like, for the example this week, somebody did respond, and they were on a conversation with a very senior person here within 18 hours.

Tooba Durraze:
Wow.

Kipp Bodnar:
That’s, like, counting sleep, you know? Like, effectively would have been, like, probably 6 hours if it, like, if the time of day had worked. So, like. I think it’s emblematic that there’s just, like, a big divide between the showers and the tellers out there in the world.

Tooba Durraze:
Yeah. Okay, so I know there are other panelists joining, so kind of my parting question to you, because again, I really enjoy how kind of in-the-know, but also practical you are. There’s a lot of hype around how the roles of a CMO are changing, and what the new CMO might look like.
I would love your take on What is still the job, and kind of what is no longer the job, or how has the job evolved?

Kipp Bodnar:
I think the job of a chief marketing officer got… like, watered down over the last 10 to 15 years, because it became about, like, managing large-scale teams, evaluating technology, and doing less marketing and less building.
And I think the biggest change is that, like, if you’re the chief marketing officer of a company, you should have Some of the best insight of who the customer is.

Tooba Durraze:
Yep.

Kipp Bodnar:
You should have the best insight of who the… of the stories you’re trying to tell. Yeah. You should have remarkable craft and understanding of, like, the way to deliver those stories, both, like, the creative and channels, and you need to be as involved in that as possible.
Like, I’ve been… I make product pages, and landing pages, and skills, and, like, all of those things, because, like, that, I think, is the role of a marketing leader in the future.
And part of, like, what that means is, like, you have to set the pace, because we’ve had… we talked a little bit about, yeah, it’s hard for people to change and evolve, but if you as a leader set the pace for, like, this is how we do things, this is the speed of which we do things, this is how we build things. Then everybody meets that pace.

Tooba Durraze:
Yeah, awesome. No, that’s great. Well, thanks so much for being here.

Kipp Bodnar:
Thanks for being here.

Tooba Durraze:
big conversation.

Julia Nimchinski:
Thank you so much. Thanks, Kep and Tuba. What’s the best way to support you?

Kipp Bodnar:
Check out HubSpot.com, check out Marketing Against the Grain. Thank you all. Appreciate it.

Julia Nimchinski:
Thank you so much. Thanks, Tuba.

Tooba Durraze:
Yeah, thank you.