HubSpot’s Kieran Flanagan on how AI is transforming SaaS go-to-market

0:28A great writing tool should give you building blocks. who've been there and done that, as they share the ins and outs of how they built their businesses, their operations, their path for securing investment, and more. Our mission with the podcast is to help you, the founder, learn how to scale your SaaS, maintain your well-being, and navigate the complexities of this ever-changing industry. I'm your host, Alex Deema, and together we'll explore the good, the bad, and the ugly in the journey to SaaS success. AI driven, ARR focused. The SaaS Stock Europe 2025 lineup means business. Featuring founders and leaders who've built the companies defining our industry, including Guy Podjani, founder and CEO at TESOL, Manny Medina, co-founder and CEO, Paid, Nicola Merksic, co-founder and CEO at Poly AI, and Alice Carlyle, Regional VP, UK &I at WIZ, to name a few.

1:29Learn the strategies, insights and frameworks that scale these businesses from startup to market leaders. Build alongside the best. Book your tickets now at sasstock-europe.com slash tickets. All right, welcome back to the SAS Revolution Show. I'm your host, Alex Themer, CEO, founder of SASstock, also a general partner of Back Future Ventures. Joined today by Kieran Flanagan, who's leading AI and GTM at HubSpot. Kieran, how are you doing? I'm doing great. Thanks for having me on. Yeah, good to have you here. This is ahead of you coming back to SaaStock Europe in Dublin, obviously where you're based and speaking at the conference on, I think it's the 14th or the 15th of October.

2:15So we're looking forward to being back at Sastoc and being back in Dublin and also seeing you there and listening to your talk. But I appreciate you coming on this morning. I think given, you know, I'm certainly following you and I'm sure a lot of our listeners are, you know, LinkedIn and, you know, with your podcast as well, putting out lots of great stuff, you know, around AI, of course, you know, at the moment. Um, you've been like with HubSpot, um, uh, you, you know, around, around 12 years, uh, sort of like on and off, mostly on, right. Um, and, um, and, um, and obviously now you are, you know, in this role focusing on AI, uh, and GTM, um, uh, and it makes a lot of sense, you know, given the era we in, is, is this the most exciting time for you in your marketing sort of career?

3:14And do you think, I guess, like for marketers in general with what AI is enabling for marketing? Yeah, I think it depends on your personality type. So I've always kind of jumped onto the latest go-to-market motion. Like when I originally joined HubSpot, it was like inbound and learning that, and that was a new thing. And, you know, in all these things, there's like the people who want the new thing and there's the people who like the old thing, right? So the inbound, there were still people who'd like their traditional way of doing marketing. Then I went to kind of lead the kind of PLG product-led growth transformation of HubSpot, that was the new thing.

3:45And that was like the new go-to-market motion. And that's like pretty prevalent now within SaaS. And so to me, like two years ago, actually, when I first used ChatDBT, the very first time I used it, I just thought everything is going to be different now. I think one of the first posts I put out was Google is in big trouble. Everyone kind of laughed at me, but it was pretty obvious to me that this was going to be pretty seismic. And so I think if you are someone who likes to figure out problems. That's really what I like to do. I don't really see myself as a marketer or anything like that. I just like to solve interests and problems and usually with new technology.

4:21And I think if you are someone who is curious and there's like all these new things you have to figure out, this is a really great time. I think it's never been, there's never been so much change. If I look at what we're doing in HubSpot in our go-to-market, we've never been running so many pilots, so many experiments, even way more than when we were kind of transitioning the company to be product-led growth. But I definitely think that there's a lot of people that are not going to enjoy this or not enjoying this because it's not very clear what kind of what your future career is. Like I think all careers are going to change quite drastically.

4:56And so it's hard to pinpoint what your career path is or how this impacts you. And so I think there's two different camps of people. I'm definitely in this is really cool, everything is changing, it's chaotic, and I like to figure out problems. Do you think like, I mean, on the career front, let's say, whether maybe it's sooner than 12 months time, that whether it's at HubSpot or other organizations, if those people that are looking for those jobs and those next roles, but they're not on board with AI, are they going to have a problem? I mean, hard to say sort of rubbing a crystal ball here, but you must have a sense of it i don't think you can be a top performer in the next 12 to 18 months unless you integrate ai into your workflow now i do think the whole ai is going to replace all humans and that's kind of been uh pushed by the you know sam altman and the fight like the core lm founders have kind of pushed that narrative and then they kind of go back and forth and they're like oh no this is kind of overblown i think gbt5 was a big wake-up call for people right everyone was kind of on the edge of the seat uh me and my good friend kip were like whatsapping each other on the youtube live and it was waiting for this incredible moment where you're like everything is different now and i don't think gpt5 delivered upon that and i think people have kind of reset expectations to say actually we need a lot of we still need a lot of people who can use ai and so like there was a a good everyone now on linkedin one of the most popular content formats is just to post whatever jobs open ai are hiring for but there's like there was an interesting one like this week where they're hiring like a content strategist right and if you actually go uh back in time a couple a month, three, four, five months.

6:30I think Sam and these other people would have pushed that AI would replace all content creators and AI could create the content. And now OpenAI themselves are hiring a content strategist. It's a good example of why you still need the human craft and the human skill set a lot of times, but that needs to be powered with AI as an assistant. I think it has to be integrated into your workflow. So I do not think you can be a top performer in the future unless you really figure out how to use AI within your day-to-day work. Yeah. And maybe a personal good example of that. I got very excited when I've seen some of these AI videos and trailers that have been played either at Super Bowl or we see them on LinkedIn with some SaaS companies that are creating them.

7:12And then people talking about VO3 and it's like, okay, I'm just going to create, I've got this idea, create this video. And then just realizing how super difficult it is. And I just spent time, you know, going down a rabbit hole and all I'm creating is like a five second clip, which is actually like not what I want. But you need a person behind that with the, you know, the video editing skills, but now AI video editing skills to actually create this. Right. So you still need the people to to use the tools. Right. Well, this is actually an important part because a lot of I spend a lot of time. I'm pretty active angel investor in precedence stage companies.

7:48And because of my profile, I see a lot of creative AI tools that can do a lot of creation for marketers or other people or other disciplines. And I'm someone who – I love VO3. I'm also trying to work on a little mini five-minute documentary only done through VO3. And the first thing I had to figure out actually is how you can keep characters across eight-second clips. right like how because what it what it tends to do is like the characters will look different because you have to replicate them each and every time across eight second clips and i figured that out but where i'm going with this is you mentioned something really important where most of these kind of tools fall down um is in it's actually the editing right so ai is ai is able to create like a one-shot version of whatever concept you have but everyone really needs to edit edit that and finesse that and ai is usually terrible at that now i know google have come out with nano nano nano banana than you you know image editing model and it is very very good but you still need the human craft to take something that ai creates and i think add like the 20 to 30 percent of magic that really differentiates it from everything else um and i think that's a pretty good example is when ai creates a content asset and you still need to do a bunch of editing to craft that magic it should be i i guess then you know with this with all these tools and with people we talk about specialization right it should then create a lot of opportunities because if people are looking for whether it's ai videos you know ai content strategist or there was a content creators like open ai uh are looking for you know at the moment those that can get on board with that now you know hopefully the future should be bright for them and a lot of demand you know for for this skill set it may become quite commonplace in you know a year or two uh but you know i guess kind for now, like if you can master, you know, a certain discipline with an AI, certain tools, probably demand is going to be, uh, you know, reasonably high, I would imagine.

9:47Yeah. There's also like the other thing that happens in all of these kind of paradigm shifts is there's a new roles that gets created, right? Like, like if you, if you think about it, some of the new roles, like off the top of my head, I was actually just looking because I was working on a, a post around this. I'll find it in a little bit, but I like some of the ones that kind of come top of the mind, AI trainer, right? That doesn't exist prior to AI. So what does an AI trainer do? Well, what's happening is you have all of these agents, you're integrated into your go-to market. So you might have an agent that kind of is really good on your website and can actually sell to website visitors and is trained to sell to those types of visitors your product.

10:24And then you have another agent that is able to come in your app and cross-sell and upsell your customers. That agent has to be retrained to do that skill set. And then you have another agent that is supporting customers. Well, that agent has to be trained on how to support your customers. So each and every time you have one of these agents, you need someone who can train the agent, gather all of the feedback, and continue to train it and refine it over time. That's an example of a brand new role that exists because of AI. And so there's going to be all of these new roles that exist in the go-to-market motion that we don't even know.

10:57And so I think that AI will disrupt some jobs, likely will make some jobs, I'm not sure, totally obsolete, but you definitely will need less people. But it will create many more new jobs. And I think that's what's going to be interesting. That's what I find fascinating is starting to see all of these new kind of roles appear because of the way AI is changing the way we work. I guess mostly over the last 12 months, over the last two years, so much hype, a lot of positivity around AI. With that, there's also some negativity, some skepticism in the market. There's a recent MIT, I think, report that 95 % of AI projects are not really producing the results that are expected.

11:44And I think that's, you know, impacted a little bit in the market. Like, what's your take on that? Yeah. And so I want to caveat this, but I haven't really looked through that report. But there were two reports. There were two reports. There was a McKinsey report. And the McKinsey report, I think, was every AI project was failing other than those implemented by McKinsey. And so that one I set aside. And then there was like the MIT one, which I think was grinding into more facts and data. Now, there was a couple of really important things about that report. And again, just from skimming through it, I think it was conducted in January 2024, like prior to reasoning models.

12:20And so that's pretty that's pretty outdated. Right. And like where we are today and where we were then is totally different. And what we've seen when I when I first kind of went back to HubSpot and we started to kind of integrate AI across the go to market. One of the kind of frameworks or mental models that I wanted to have for everyone was we're going to we're going to go after the most transformational use cases that we believe will be the most impactful in the future. even if the models cannot fulfill that need today, but we're going to build the organizational rigor, the infrastructure, and start to integrate that into our go-to-market motion.

12:52And when we get updates in the models and put the models live, all we have to do is plug them in, we get that benefit. That has proved to be true, that some of the things that we did, that we were in pilot mode, they didn't work that well, but we had all of the work done and we could plug the models in. We're starting to see huge gains over the past six months in particular. So when you look at that report, it's kind of outdated in terms of how capable the models are the other thing that really is important about that report and this is the part i wanted i had on my list to dig into this week is it's the adoption part right so a lot of the problems isn't just there's like two other problems i would point out it's like number one is adoption getting people to use the ai functionality you're either bringing in through vendors or you're piloting is like a really interesting thing because in a lot of cases people see ai as a means to displace them and so sure the leader is like really excited i'm going to bring in ai and make this team much more productive but when you go all the way down the chain for the people who have to use the ai and actually integrate it into the workflow it's first of all it's hard to get anyone to change their behavior user behavior is one of the hardest things to shift and then the second thing is if you're using that you're like am i just training this ai model to do my work and in some cases yes it's probably true right so like adoption is a really uh complex problem to solve and then the third one is there's like an interesting thing do you do you bring in vendors or do you build a lot of custom ai functionality yourself and that's a hard thing to figure out one of the problems with bringing in vendors is a lot of these companies are wrappers like ux wrappers around prompts and because the prompt is trying to serve many many companies it's kind of pretty generic right it's it doesn't it's so it doesn't do a great job whereas if you actually customize this you as a company have a ton of nuance maybe your outreach tool maybe whatever tool you're using there's a special kind of way that you want to do it so you want to refine that prompt and that's a that's a reason that you would actually end up customizing a lot of the software yourself versus buying something off the shelf but customizing it adds a huge burden onto the company you have to build that you have to maintain it you have to make it as good as what's available in the market.

15:01So there's like some other interesting issues that arise would bring an AI into the company that isn't just, is the AI capable enough, like the model capable enough? And I think the takeaway everyone took from the MIT study was, well, AI is just not good enough. And I think there's just more complexity in terms of making AI work within the company than just the model's capabilities. Can you give an example then of what's produced results for you? something you've done specifically within HubSpot that's worked well for you and what you can share around that? Yeah. So we've integrated into, again, I think the interesting thing I'm going to give you is there's going to be no surprises, like no kind of hidden surprises here in that we've really integrated data as a big part of it, right?

15:45So I'm going to start with data because the context layer, your data layer really matters. like your data hygiene, the quality of your data, the quality of your data sources, they really will dictate a lot of the success you have around AI. So one of the examples would be on our website, we were able to integrate AI into a lot of the ways that you interact with HubSpot. And when people think about that, they're like, ah, that's a terrible experience. But you kind of remember the chatbot era, right? With that chatbot workflow era. And it was kind of like very, it was like a far worse experience than like interacting with a human.

16:20AI is much, much better. And so when we look in the back end, we can see the way people feel about interacting with AI is not too dissimilar to how they feel about interacting with a human. And that is an example where you have to have great context because when someone joins and they start to chat with your AI, if you have context about where they came from, where they visited, any kind of stuff you have in your CRM, you can have a much more personalized conversation. So we have an AI bot that's able to do a ton of support, deflection, and then also able to book meetings for the sales team. So they actually can do sales conversations.

16:54Personalization has worked great for us. We've integrated AI through all of the personalization, prospect and email workflows that we do. We've seen like huge, I'm surprised by some of the gains. So we have a ton of emails that go out to prospect to different types of customers. We say our prospects and customers, we segment them all these kind of different ways by fit and intent. And then we use AI to like really personalize the messaging. And we've seen anywhere between like 300 % and 500 % increase in booking meetings, right? Just by, again, data quality, really perfecting the prompts. Like I want to give the team a ton of credit.

17:29This is one of the things I will say that I maybe got wrong is I've always been on the, you have to prove something with a six-week cycle. And if you can't prove it, you should move on to something else. Whereas the prompts that really started to work in terms of personalization took us like three to six months. So there's some things, again, you should just take a bet on and get rigor and discipline around. that's worked really well. We have some other interesting experiments, I think, that are probably in the kind of real bleeding edge where we have a digital avatar that is starting to do first kind of interaction with prospects.

18:02The kind of first sales conversation you have is with a digital avatar, and that avatar is trained to actually do a bunch of discovery qualification and give you all the answers before you actually interact with a salesperson. What's really interesting about that is when I stood that pilot up, I was like, is consumer behavior really ready for a digital avatar? Are people going to think that's really weird? And the average engagement rate was eight minutes. We had someone that talked to the avatar for two hours. I don't know what they were talking to the avatar for two hours about, but the engagement rate was through the roof.

18:33It worked really well. And over time, we saw better closed rates. And so there are some examples. I can go through many more, but I want to pause and see if you have any questions. But I can go through some others. I mean, just curious, the digital avatar, is this an internally built tool? Or is this something that you guys have trialed that is out there in the market? And can you share? Yeah, we're using a vendor called OneMind. Give them a shout. Yeah, you've probably heard them. Amanda's CEO, great CEO. So yeah, we have a great relationship with OneMind. They've helped us to build a bunch of custom use cases.

19:08and again it's an example of we probably started using OneMind pretty early in terms of where consumer behavior is but what we want to do is find the signal and over time we believe the technology and the consumer behavior will catch up I was surprised by how engaged people were off the bat with the avatar and again the avatar you can do a photorealistic version you can just do the orb that you don't need the photovoltaic version. But I do think one of the bets that I have a concerted belief that will transform the way B2B, kind of just B2B go-to-markets work, is multimodal, which is where the agent can kind of talk to you, hear what you're saying, do video, and then see your screen.

19:53And I think that fundamentally changes the way people are going to buy software. And so we have a lot of multimodal pilots up and running to see where it can come to fruition across the go-to-market. Well, I guess with the AI tools, some of them are relatively inexpensive to use, but also if you start to use a lot of AI tools, all of a sudden your stack is becoming pretty big. You're running a lot of pilots. It can be quite expensive, right? And so consideration around budget, especially if you're a smaller company, is something that our listeners kind of need to think about. HubSpot, obviously a much bigger company, than I think a lot of our listeners who are a bit earlier stage.

20:38What do you think is something that, you know, our listeners could be doing around, or should be doing around AI with GTM that you're also doing or what you're seeing in the market? Yeah, and I think one of the things we try to do is because there's still a lot of noise to signal in AI, which is like, you can stand up a ton of pilots and only one of them will work. and that's an expensive way to try to find, you know, signal within the noise. And so the reason we run so many pilots is because we have a go-to-market platform and so we're trying to find the AI use cases that work that we can then integrate into the platform and that's kind of what we've been doing.

21:19And so our go-to-market runs ahead of the platform and so the platform can integrate things that we can bring to smaller customers and we can be sure that works. And the reason I'm saying that is because if I'm a smaller customer, I would probably just bet on the things that I know are working. and I have enough signal externally to tell me that these things will work. And so, again, I can't emphasize enough that if you have bad data, it doesn't really matter what I'm going to tell you because your AI is probably not going to produce great results. So I think trying to solve the data layer, trying to make sure you have a good context layer, which means making it easy to load context into the different AI initiatives you're standing up.

21:58The personalization one really works. Wherever you want to bring it to life across your go-to-market motion. But again, it's not, this gets talked about a lot externally, AI to be able to do kind of what the SDR does or the BDR does, so warm outreach, cold outreach. I think cold outreach is really difficult now because it's so saturated with AI. What I would say is the more creative you can be with data and don't just follow the same sort of outreach templates that everyone's using, you can still get some pretty large gains, but you have to be much more creative with your kind of data use. I think AI for support deflection really works.

22:35Again, not groundbreaking here. One of the ways that we've integrated AI, if you have a small sales team, is like AI to enable sales to do more selling versus kind of sales admin. So like call prep, deal summaries, the ability to actually preload the CRM record with like core talking points. So what we do, a bunch of regression analysis and we can say, hey, like this customer, or this prospect is very similar to this customer that you closed. And these were the core selling points that really resonated with that customer. And so when the sales rep gets on the call, they actually have all of that preloaded in front of them so they can do a much more consultative sales call.

23:15And again, that's not, for a smaller company, that is possible, whether you use an off-the-shelf tool that can do some of this for you or you kind of just build some custom internal tools that can do that for you. And so enabling your sales team to do more consultative selling really works if you take that unstructured data and turn it into insights for the sales team. There are some of the ones. I think if you were a smaller company, maybe what you spend your time on is trying to, because there's two things we're talking about here. That is how you integrate AI into the core go-to-market, but then it's how you integrate AI into your go-to-market team's workflows, right?

23:48So how I can just be a better go-to-market professional because I use AI day in, day out. And I think if I was a smaller company, I might start there, right? I might say, hey, here's a seat of whatever, Chats PD, Gemini, Claude, I really want to start to integrate this into the way you work so I can get 20%, 30 % more productivity of the team I already have. What are your thoughts around AI and content marketing in terms, maybe specifically around, let's say, on LinkedIn right now, there are a lot of CEOs, there are a lot of founders, but not just, right? Other roles that are using posting regularly on LinkedIn, building out great audiences and using that as kind of top of funnel for their business.

24:36One example, Adam Robinson, you've probably seen a lot of his content over the last couple of years. I would have said, and feel free to challenge me on this, but I would have said or I felt a couple of months ago that if you'd used AI to create a LinkedIn post, that people would have noticed it. it wouldn't have got the right engagement. And maybe there would have been a little bit of like, because they noticed it, pushback, like people saying, this is written by AI. This wasn't written by Kieran Flanagan or Alex Sumer or Adam Robinson. What I actually noticed this week, I think whether it was even sort of yesterday, like for instance, Adam Robinson is openly talking about he's just getting AI, Claude, I think it was, to write like Adam Robinson, his LinkedIn posts, where he probably spent a lot of time each week prior to that creating this.

25:38And now he can spend more time as a CEO on the product. So what are your thoughts about what's changed there? Is it just got better? It's got more acceptable over time? And should everybody be doing it? And then is everything going to be written by AI? Will that affect engagement on LinkedIn? Yeah. Okay, so this is one of the places I've spent a lot of time because a year ago I was teaching people how to really use AI in a novel way to do short-form content. And it was basically what I was trying to show is you can build kind of write-in style guides and then you can use that to differentiate your content so it does not look like it's generated through AI.

26:21And so even a year ago, if you did that through Claude, it would have been hard to spot something was created with AI. I think the problem is a lot of people, They just go in with a, you know, a basic prompt. They ask the LLM to create something. And it's pretty, it's so easy to tell that that was created by AI. Now, what happened to me personally, and I'm not saying that this, you know, my reach is big enough that this influenced lots of people on LinkedIn, but we did a pretty popular show with Dharmesh for Marketing Against the Grain. And what was really interesting, I think that episode has like 200 ,000, 250 views between RSS and YouTube.

26:56And I give away like the actual method to do it. Like I wasn't giving it away because it was like, oh, well, this is like some of my secret sauce. And I give it away because we wanted to do that on the show. And what I what was really interesting to me was how quickly my my different writing styles were replicated in LinkedIn. Like I literally could see people had taken that and it was like really easy to spot. So I was like, OK, well, this stuff is like the challenges when one person does it or, you know, when you're doing it, it's like hard to spot. And then when you have everyone doing it, it's really easy to spot.

27:27And AI is able to templatize things so much faster. And so your tactics spread so much more rapidly. Now, the models have got a lot better. I launched a small little writing tool as a test, took that down. And I have a far bigger writing tool coming out actually in the next month, which is specifically for actual founders and people who want to create content. And the models are much, much better. Like they actually are able to create a bunch of content that is better than most people. But I would still advocate for this. Like this is my core thesis. Even for someone who's going to use my product, I would never copy and paste content from AI.

28:07For two core reasons. I still think it is easy to spot actually. Maybe it's just because I spend so much time in AI for content creation and I've created software around it. But I think it's very easy to spot. And then the second thing is there was another great study, which I think is one of the more important studies around AI, of how AI makes you dumber, for better words. If you continue to outsource the AI, you lose part of your skill or part of your craft. And I think writing and content is going to be one of the most important skills you can have because everything is going to get noisy.

28:42people will gravitate towards people which is why founder content is working so well because people people don't even really want the brand content anymore they want the human-led content which is like starts with the founder and so i think content is a really important craft to to master and outsourcing that to ai to copy and paste feels like you will diminish your ability to like really do that over time that should just in this case does the founder need to create no they can have like an internal ghostwriter who can like use AI as an assistant, but I would not, I would never copy and paste content from AI.

29:16I'm not, I'm not like I do, I think Adam is an incredible LinkedIn creator. So this isn't to do with what he's doing. But I just think if on mass people start to use AI as a core like content creator, it will, it just will be easy to spot because AI will create content in the same way that for you as it creates for me, as it creates for everyone. Like it's all gonna look very similar. Yeah, no, important points there, I think, because, like, you know, as a founder myself, you know, constantly trying to get better at writing, you know, to do more because, you know, get the right strategy and cadence and find, like, my North Star.

29:56You know, writing, it's a skill, you know, it's, I guess, kind of, like, highlights or indicates, you know, a level of, like, intelligence as well, like, develops you. and then as you say and I hadn't really thought about it sort of too much that if you start shifting these sort of skills to AI to save time to be more productive then it's it's clearly going to be you know impact and development and then if I don't know a year or two's time you have to get back to writing yourself you're like you're gonna be a bit of a block right like oh shit you know I'm really out of practice and perhaps it's maybe akin to using the calculator to do maths and simple equations instead of thinking about it in your brain and just kind of looking for the easy route, right?