AI is NOT What You Think: Unlocking Growth in a Chaotic Era!

everyone to be in business, for you to be a CEO. So, I'm really looking forward to uh spending a bit of time getting an understanding what's going on in your head and how you're going about dealing with what everybody else here is uh Yeah. >> facing into. It is a very fascinating time indeed. We were talking backstage about how this is all about change and managing change, and you've done that significantly throughout your career. So, I'm looking forward to this conversation as well. It is all about change. Let's kick off with the big picture. All right. >> Cuz we do live in, I don't know, call them whatever you want, unpredictable, crazy times. The 2020s, for sure, is the craziest decade I've experienced over many. Um AI now being the accelerant and the thing everybody is focused on is a significant shift. So, how do you see things in the state of the tech industry and more broadly? I know. If you say that this is crazy, I will take it because I joined HubSpot January 2020. A little bit of didn't know what was going to happen, but a lot has happened. And I would say that uh you know, beyond the COVID years and dealing with all of that, uh 2022 and uh specifically November 2022, when ChatGPT came, there was a big shift. And you know, you and I have seen this. It's a platform shift. It feels overwhelming. It feels like things are changing every single day. I do have a couple of slides I want to share, you know, just kind of like, you know, walk us through where we are. This is what happens during a platform shift.

And that's what we're going through in the industry. Every 15 to 20 years, we go through a platform shift. And when we go through a platform shift, a new technology emerges. People start building, many times over building. And then there is a question of who controls that technology. And then there's a question of who wins and who loses in that new technology wave. It's as simple as that. We've now seen it through PC. We've seen it through internet. We've seen this through mobile. We've seen this through social. And the same pattern follows, right? So, if you take internet, when internet emerged was completely new as a technology, and you saw level of overbuilding and building, which is kind of happening right now, right? Back then it was, you know, routers, vans, lans. And I know you remember this at um there was a point in the early 2000s where 97% of the fiber that had been built was not used. It was called black Yeah. >> fiber. That, you know, was a level of overbuilding that happened. And then the new technology emerged. Google became one of the gatekeepers of search and new business models uh like DoorDash, like Uber couldn't have emerged without the combination of internet and mobile. And uh new winners and and losers emerged. I think the same thing is happening with generative AI. Obviously, it's a new moment, and there's a lot of building and I would say overbuilding that is happening in terms of the data center capacity. And that is going to lead us to a place where there will be new winners and there will be some losers who are not moving quick enough. I think that that's the same pattern, Steve, but my observation is two things. One is

that it's happening much faster than any of the previous cycles. What we saw happen over 8 years, 10 years is now happening over 3 to 5 year period. And uh everybody's got a megaphone. And everybody is on a podcast. Everybody is on, you know, X. Everybody is on LinkedIn. Everybody's got an opinion. And so, it feels even more chaotic to navigate through what we've seen as a platform shift. >> Yeah, absolutely. And it's there for us all to see and use, too. Yep. >> And once you immerse yourself in it, you you can see the, in a sense, the the scaling up of of the way you're going to interact with these new tools, let alone how your organization is going to respond. Absolutely. Absolutely. Yep. So, just talking about um people in the room and how they're responding, how do you see the response so far by Australian customers? And should they be thinking or well, how should they be thinking about their investment in AI and the outcomes they should uh want for it? I first of all

love the way you framed it, which is what should be what should they be doing for the outcomes? Because it is about, you know, technology driving outcomes, not technology for the sake of technology. But let me ask you this, like I can see most of the room. Um how many of you use AI every single day? Okay. How many of your organizations have seen significant growth and outcomes because of AI? Okay. Okay. So, about like, you know, half. Uh I would say everybody is using AI, whether it is an LLM or some kind of tools. And there's a slightly lesser number or percent that has seen organizational transformation because of AI. That is the state of adoption, and that is the state of transformation right now globally, right? And you asked specifically about Australian customers, but I think globally what we're seeing is that everybody is using AI. Everybody is, you know, got at least a couple of LLMs, a combination of some tools in their daily usage. But we're all trying to go from what I call individual AI

productivity to institutional AI productivity. And that means individual AI productivity is like, you know, you getting up in the morning and you're writing an email, and now you have a better thought partner, uh maybe some, you know, LLM that actually understands you, your style, and can give you a better email. And then there is institutional AI productivity, which actually requires a lot more context. Because, you know, the the change between individual and institutional is you you need to get the growth context of the whole organization. Your best campaign manager, how do they pick the best campaign? Your best sales rep, how do they think about the best deal to work or how to save a deal that is not moving forward? Or your best support person and what should they care about? That requires a growth context. And so, I think that's the stage that we're in. I will say that specifically calling out some Australian customers. I know some of uh many of them are in the room, but there there is good adoption and movement towards this. You know, Hungry Hungry is one of the customers. I hope you're there in the room. Um very interesting in in how they have completely reimagined marketing with Loop, which is the playbook that we have been talking about for a while, as well as, you know, customer agent. They're resolving tickets using that. Um Cannibuild is another Australian customer. Again, leveraging AI in almost every aspect of their go-to-market and going beyond the experimentation stage to scaling with it. And my my own favorite is Builderness, New Zealand-based company and beautiful country. And they're actually in the hospitality as well as, you know, uh helping helping folks kind of figure out how to uh spend time in New Zealand. And they are also adopting AI in almost every part of the go-to-market. And I think that we're in that process of translating real productivity gains into go-to-market. And that's a, you know, it's a fascinating time to be here. It's interesting when you talk about those organizations. I would pretty well guess, without knowing many of them, that they're well led. And what I mean by that is the job

we all have as leaders to get the institutional benefits of AI is a leadership challenge. It's about cross-functionality. It's about being explicit from the top down as to what it means to be successful and what actions are expected across the organization. So, I think that's the the challenge, cuz it is much easier to adopt individually than collectively. Yep. >> Um >> Um actually, we should going on to something else I think is really important here is just some of the best use cases that you see that are driving results. And And how should organizations that are beginning the journey, how should they sort of tackle it? The answer is what are your

company's goals? What are your team's goals? Because while AI is a new platform shift, and of course, you know, it is uh it is something that everybody wants to have a roadmap. It has to start with your company, your goals, what you actually want to reimagine, and not the other way around. So, I think almost every customer or prospect conversation that we have, we start with what are you trying to drive? Where have you seen the biggest challenges in terms of how you want to grow? Let's then identify the use cases of what that looks like. So, it really has to start with you identifying the biggest challenges in the way like, you know, it's new ways to solve old problems. That's what AI is all about. But having said that, we are now 3 and 1/2 years into AI. And there are a set of very specific use cases that can help you unlock growth. And now I wanted to share and I'm going to like flip to what we see as AI customer journey. And again, you wouldn't just start with the entire journey. You'd actually pick one or two areas that makes the most sense to your business and what you want to solve. And let's start with AIO. I would also plug Asia Frost who is, you know, speaking today. I would really plug her whole session because she started experimenting with AIO two and a half years ago for HubSpot. And what we found is that there

is a completely new way in which you need to show up in the answers that LLMs are providing, right? We've we've made it a science to show up in Google searches and blue links and how to show up in blue links actually came up. But then now almost 60% of Google searches do not end in a click, which means you have to find diversified channels and AIO and showing up in LLM is one and that's a place that, you know, we we really encourage people to think about and, you know, Asia will talk a lot about that. Um, we're also finding, you know, talking about like old, you know, problems that we've had, uh, it's pretty

difficult to understand who your ideal customer profile is and how to reach them at scale. And so a lot of our customers who are trying to kind of like expand their growth, they start with the data agent and prospecting agent. And in super simple terms, data agent will enrich all the companies that you have within your CRM and the contacts that you have within your CRM and it will get the latest information so that you have an ideal customer profile matched TAM so that you can expand there. And then what we are finding is prospecting. Steve, I don't know when you started in sales I started in sales, the difficulty was the first time I got I had like 500 accounts and it was pretty hard to cover 500 accounts. You're going to go and look at like 10 case and 10 Qs and actually find information about every single account. Now you can do it with AI. That's the thing that AI has made super easy and so we find that prospecting agent you you can actually identify the right intent signals and then from there you are able to send personalized email based on your specific company and your specific you know, problem that you're trying to solve and that means better open rates. And so I think like in the sales arena, we are beginning to see that. And then, you know, again talking about old problems with new solutions, you know, in sales the hardest thing was trying to take notes during a call, entering that information into a CRM and then doing a follow-up. You just didn't have enough hours in the day to be able to do that. Now, what you can do is just have a bot listen to your conversation and tell you what the action items are, even give you an email to send and that makes it much much better for reps to focus their time and energy. And so I think those are really the the top use cases that we're beginning to see from a sales perspective. And then the last

thing I'll finish off with is customer agent which is support. In fact, like I think one of the first use cases which found product market fit is the customer agent where you can resolve questions using your knowledge base, your history of how you've responded to your customers and we now have over 10,000 customers at 70% resolution rate, um, you know, resolving tickets and therefore scaling using AI. So some of the handful of use cases, but my my, uh, you know, guidance to any one of our customers and prospects is basically start with your problems and what you are trying to solve where you have challenges and then pick an area and then go deep in that area and once you scale that and once you get confidence in that, then you can pick other use cases. This is awesome. And this is also, in a sense, about, um, you know, raising the bar on the quality of your sales and go-to-market effort. >> Yes. >> And I mean, we can all see that from our day-to-day experience how the sales experience we have are not as good as they could be. And the connection there to growth is is just undeniable. So >> it's wonderful. Good stuff. >> Yeah. Um, in terms of go-to-market org structures, how do you see them changing

and what does that mean for the people that, uh, leaders should be, um, looking to bring on board? Yeah, I I think, you know, that that it's a completely different aspect and glad we're talking about that. Um, so what is what what is interesting, Steve, is that when uh, internet era came, what happened is that all of your buyers were searching for information and therefore what happened is that marketing and sales started to blur. Yeah. Because every single buyer of yours, they search for information, they already got information and by the time they actually came to a salesperson, they had read, you know, enough about you on the internet. And what that meant was that the line between marketing and sales started to blur. Marketers needed to get much more personal and personalized in terms of the information that they were sharing and sales people needed to almost act like marketers because the lines were blurring. And that happened in the internet age. I think what is happening right now and what is very interesting is that sales and customer success, the lines are blurring. Mhm. And within within the go-to-market organization, that's because, you know, with AI, you want to be able to get the value and immediately expand the value of AI and it's almost matter of minutes that you really need to see the value of an AI use case or an AI agent and that needs to expand, which means the handoff between your sales team and your customer success team, it first of all cannot even be a handoff. It just needs to be this continuous process of helping customers get value. And so I think the more fundamental change that's happening with the AI era is CS and sales, the handoffs are really blurring within the go-to-market organization. One of the things that at HubSpot that we did, you know, coming into this year is we kind of moved sales and CS into the same org and we want that team to move faster, be able to present ideas during the sales cycle to customers and then make sure that the value is delivered really quickly during CS. And so I think that over the next two to three years we're going to see that handoff change. But going to maybe the question which is what does that mean for leaders and how should they be thinking about hiring?

Um, it's it's a pretty transformative time and what we are finding is that as these lines are blurring, it's much more about having people that are curious, that are leaning forward into change versus fearing, you know, the change itself. And they're also builders that that are experimenting with the technology and learning with the technology. And you and I were talking about this in terms of how leading through a time like this much more about navigating change. And so I think like, you know, what we think about as curiosity, experimentation, learning mindset, being a learn-it-all, those are the kinds of things that we need to hire for. >> Absolutely. Absolutely. It's the you're talking about high-growth culture Yeah. and, and and a mindset for change. >> Yes. Which, you know, curiosity goes right to the heart of it. The other thing I think for all of us, it's a big thing at the center of this is our ability to, one, to demonstrate learning at the highest level, I think, which is seeking feedback from those around you. And also being open to what you don't want to hear. >> Yeah. Is a really a really important element of that, particularly as you're going through these significant organizational uh, changes or trying to deal with things that, um, nobody knows the answer to. So >> Yes. mistakes are going to be made. How you and how others going to react given that mistakes actually are the journey. Make no mistake about that. Yeah. That is the journey. So how we react and respond and how we treat each other, you know, within you know, customer success, in sales, it's really going to come to the fore over the next few years and it's something everything we're talking about is something we've been looking forward to for for many, many years. You know, how do you get, um, customer success and sales to work better together and and, you know, it's it's, uh, it's something now that, uh, is real and a real opportunity for us given what AI is providing. Absolutely. And I I like what you said, mistakes are the journey. And, uh, you know, one of the things that we've encouraged our teams to do from a mindset perspective is really, uh, the experimentation mindset and start with a hypothesis. Prove yourself right or wrong within that hypothesis. It doesn't mean you're wrong. It means the hypothesis was right or the hypothesis was wrong. And encouraging that and being, you know, almost celebrating when you prove something wrong Yeah. as much as you prove something right is the kinds of mindset that we need and that's a big shift as we navigate through, uh, this this platform shift. >> Sure is. We might talk a bit more about that later. Yeah. Uh, I want to move to something which is, uh, happening out there which is people like me starting to code using vibe coding and this change that, uh, is happening where you know, we think to the end of the engineer is near and all this crazy stuff that, uh, people are talking about. But, um, do you think that this, uh, shift or the the fact that AI makes it

easier to write software, build products. What do you think it means for this whole question of build versus buy that organizations are going to have? And where do you think a platform still wins today? I love this question. I mean, everybody is vibe coding on the weekend, right? You know, everybody is vibe coding. And so, the whole conversation that's happening within the industry is, you know, coding has gotten easier. So, where are you going to apply that? And you've been in the application space for a very long time. And, you know, at HubSpot, we've lived in this for 20 years. We have 3,000 engineers who live and breathe in, you know, agentic coding. And we've not vibe coded one core application. Because one part of it is writing code, but the rest of it is maintaining that code. It is integrating it with, you know, 20 to 50 other applications that we have. It is making sure that you're standing with the best practices of today. You know, if something is changing in AEO, you want to have that reflect within your, you know, solution, the agentic solution. So, I think there's a lot of difference between your ability to code versus your ability to kind of have a scalable system. And so, I do think that the build versus buy debate, the core of your, you know, organization's growth and maybe, you know, couple other areas, you know, you're not going to vibe code your payroll system. You're not going to vibe code your supply chain in, you know, within within the core ERP systems. And, you know, those areas you're going to like continue to buy. And then, on top of that, you probably will drive agentic automation. And that's what we're beginning to see. And one of the key things that from a HubSpot's perspective we're seeing is that also, AI output is very different from AI outcomes that you can drive. And in order for you to drive output, think of an output as, you know, a blog post that you can write or an email that you can write. Now, does that output actually drive growth? Yeah. Now, in order to do that, you actually need your history of, you know, emails that you have sent to your customers. Which competitive value proposition wins and which ones do not. How to represent your brand. That is the context. And when you give more of that context, then the output translates into growth outcome. If you have a very clear prospecting email that knows your company, knows your brand voice, knows how to represent you in addition to how you compete and win against your competitor, and you can send an email, that email gets a much better open rate and therefore can help you win more deals versus an email that you get generically. And so, there's a lot of difference between, you know, just a generic email versus the ones that can actually drive outcomes. And so, that's how we think about it. There is a place for vibe coding, but it is much more of taking the context that you have within your organization and growing on top of it. And then, there's a place for, you know, overall set of applications, you know. And you'll be faced with lots of people coming to you saying, "Look, we can do this. We can do that." And I always say there's a a great way to follow up an idea, which is to say, "So what? Now what?" Yep. Which is, okay, so you can code this. So, so now what? Where does that lead us? And I had that experience myself with something I just knocked up. And I just sort of I looked I looked at it and go, "Well, this is great that I've done this, but where do I go from here?" Yeah. So, Yeah. I think that's a really important point is that there's a lot more to the context and how it really plays into these ideas will play into the most important things for your organization. And talking about organization, let's talk about HubSpot. You are and have been leading through a massive internal change. I think you told me that there's nothing that hasn't changed and given what is

going on. So, what are you changing? And what are some of the, let's say, the most significant lessons you've learned in leading that change? The answer to the question is like, we've What is it that we have not changed in the last 2 and 1/2 years? And I do mean it because we talked about the November 2022 moment where ChatGPT came in. We had a plan for 2023 and we had just set that plan in motion as we came into January 2023. And we said, "Well, it doesn't work anymore because there's a big platform shift." And the question that we asked ourselves as a leadership team was, "Do we think that this change is incremental or do we think that it is a step change?" And the answer even then, even when the capabilities were not as well developed as it is today, it was very clear that it is a platform shift. And therefore, the first thing that we started with was actually changing how we build our products. And that was a huge change. And I'll kind of like, you know, maybe if this is helpful, walk through how we build has transformed. And the first stage for us in terms of building the product was um co-pilots. And we know back in 2023, it was co-pilot and assisted coding. And we started with using GitHub co-pilot. And we were leveraging a co-pilot for a little while. And what we found that the reliability of our product was not impacted with assisted coding. And so, it gave us the confidence from an organizational perspective. And it gave us the data that we could continue leveraging that. And then, what happened in 2024 is that agentic coding just became incredibly good. And we saw a step change in the capabilities of the models. And so, for us, the second stage was really leveraging cursor, cursor, leveraging Claude code for agentic coding. And we found, you know, fairly big improvements in terms of engineering velocity, in terms of engineering productivity. In every one of the metrics that we were tracking, it was like pretty big step change in terms of how we were building. And then, what happened was that all of these off-the-shelf coding agents, they were not optimized for the HubSpot developer environment, which means the very specific libraries, the very specific ways in which we build and what our developers need to access was not available. And therefore, we built an entire infrastructure for a containerized environment that we could optimize specifically for the HubSpot developer. And that was the phase three. And when we did that, it is now where all of our agents speak the same language. They access the same data. They access the same tools. And they are able to get to the same growth context that we have built. And that has really enabled like 100% of our developers are building using our, you know, technology on top of the agentic coding. And that has meant that our pace of innovation has been really, really, you know, accelerating over the past couple of years. In addition to that, we've changed how we grow, right? And, you know, I talked about Asia, you know, and and she's fantastic. Our marketing organization is in fantastic. And one of the things that we started in terms of our own growth is actually a set of agents and assistants that we built internally, right? And the first the set are demand agent. So, the demand agent actually for us, it builds up our ICP and it expands the enriched contacts that we can actually access. Then, we started with an inbound agent. And so, if someone comes to our website today, 82% of the questions on our website are handled by this inbound agent. We've given it our pricing information, campaign information, how to qualify, how to provide competitive information to our prospects and customers. And so, a lot of that has transformed. And, you know, AEO, which we talked about a little bit before. So, the top of the funnel and how we think about marketing has now become completely agent first. And then, when you go into the sales cycle, prospecting agent obviously allows for a much more personalized outreach. And we started using that a couple of years ago. And that meant our BDR productivity, our rep productivity has improved pretty significantly. And then, you know, what we also found is that our reps, they like having an assistant. They like to have a conversation to be like, "Tell me what the risk of this deal is. Tell me if you know, what I need to be looking at in terms of closing it this quarter versus next quarter. What is the stage the deal is where I should be giving, you know, much better examples and support for for the deals." And so, we now have those types of assistants in the sales process. And we found that the combination of both the prospecting agent and assistants in the sales process is improving our win rates and productivity. So, on that, we've seen a lot of change. And then, in terms of our support, 60% of our support tickets are handled by customer agent. And what we found is that, you know, we could probably bump it up more, but we don't want to sacrifice the customer, you know, satisfaction. And so, we've really balanced what that looks like. So, you know, again, case in point, almost every single aspect of how we build our product and how we continue to grow has changed with AI. And that is also because of how our teams have really embraced and experimented with AI internally. And what what's the look, if you say what's the one thing in terms of your experience of going through all this that you'd want to share with everybody? It's really about having a

you know explorer mindset. You know, you said this earlier mistakes are the journey and I think that resonates deeply with me. You know, one of the things that we've been talking about overall within the company is that in in the previous decade it was almost like a map. You had point A, point B, there were known playbooks, there were known place that you could take to you know grow through that journey, but right now it is experimenting, it's learning, it's making mistakes, it's you know, pivoting very quickly and being very agile as an organization and as a leadership to go through that and that has been the biggest part of the journey and it's it's that's also the more fulfilling part. You know, it's a bit freeing to say nobody knows the answers and therefore you can only learn by doing it and by experimenting and the faster you learn as an organization, the more you're going to you know get the results and so that has been part of the journey. Yeah and look I I really believe that individually and collectively our is by far the most important thing you need to be thinking about and developing in your organization and it requires all those attributes that Yamini just mentioned. So it's a big leadership journey that we are all on. Yamini, thank you so much for being here and sharing your experiences with us. It's fascinating, it's exciting to see where your organization is going. Please join me in thanking Yamini. Thank you so much.