SaaStr 836: The Step-By-Step Playbook for Building AI-Powered GTM Teams with Personio's CRO

0:01Welcome to the official SASTR podcast where you can hear some of the best SASTR speakers. This is what the cloud means. Up today on the Sastr podcast. Let's talk about another assistance that we built. This is for our expansion SDR. And our expansion SDR does cross-sell all new products into our existing customer base. We have about 15 ,000 customers. So we're looking to cross-sell into them. The problem here was that when we did our jobs to be done mapping, that we found that every expansion SDR was spending two hours a day finding customer information to really make a relevant call. And we said, okay, let's change that.

0:50And one of the go-to-market engineers built like this assistant. Here you see, for example, that the research time that an eSDR spends on this work went from two hours a day to 15 minutes.

1:09Hey, Sasser, imagine having agents for every support tab. One that triages tickets, another that catches duplicates, one that spots churn risk. That'd be pretty amazing, right? HappyFox just made it real with Autopilot. These pre-built AI agents deploy in about 60 seconds and run for as low as 2 cents per successful action. All of it sits inside the HappyFox Omni-Channel AI First support stack. Chatbot, Co-Pilot, and Autopilot working as one. Check them out at happyfox.com slash Saster.

1:41Hey, everybody. Saster Annual will be back May 2026, the world's largest SaaS and AI gathering for executives. Just as last May, we hosted 10 ,000 attendees with 68 VP level and above attendees, 36 % CEOs and founders, and 25 % were AI first professionals. It's the very best of S-tier attendees and decision makers that come to Saster Annual and AI Summit each and every year. But here's the reality, folks. The longer you wait, the higher ticket prices get. They're cheap now. They're cheap, so just get them. Early lock in your spot today. Use my code Jason100 for exclusive savings. Get your tickets at podcast.sasterannual.com or just use code Jason100 when you check out.

2:22See you there. Saster Annual and AI Summit 2026. It will rock.

2:30Hi, everybody. Welcome back. My name is Philippe Lacour. I'm the CRO for Presonio. We are a late-stage HR and payroll platform. We have about 1500 people and Munich headquarters. I will talk today about our journey to become an AI-powered go-to-market. Back in May, we had a big AI week. with our company. We called it the AI Search Week. We gave all our people access to LLMs. We had speakers from OpenAI, Mistral, AWS. Our CEO, Hanno, and co-founder kicked off the week. And then we had project teams who could build, agents who can build with the help of Leica engineers. And it was a huge success.

3:26The entire company was buzzing. And after that, we've been monitoring usage of LLMs and AI in the company. My go-to-market team uses now about 90 % of the people use LLMs every week. And so this was a big success. However, after the AI search week, we felt that although usage was high, that this was maybe not enough to reach true transformation and to really fundamentally change the way we go to market. And so we started an initiative that is called AI Powered Go-To-Market. And this is our Slack channel, which we started about mid-June, so about six weeks after the AI search week. And this was started by myself with the help of a lot of people.

4:22And what I will do is present five lessons learned that we have so far and the journey continues. And then I'll cover like four use cases. And after that, we'll do some Q &A with Amelia. So five lessons learned in building the AI-powered go-to-market. Lesson number one, next to like a great bottom-up motion where you give all your people access to the tools and where you train them and where you help them, you also need a top-down motion. And the top-down is important because when you start talking about real transformation of your go-to-market, you need to make decisions about resource allocation.

5:07You need to give people permission to spend a lot of time on changing workflows and changing ways of working. You need budget. We heard earlier today, like, okay, you can start testing, like, lots and lots of tools, but ultimately you need to decide which tools to go for. So I think it's very important that in the journey from experimentation to scale, and a lot of companies write like, hey, I'm staying in this experimentation phase. How do I now really go to scale? That you not only have a bottoms up motion, but you also have like a top down decision making. Lesson number two, the cross functional approach is really vital.

5:50and we have we're lucky to have a data and systems team they own our infrastructure our systems our snowflake instance we then have like revenue operations and in revenue operations we now have two go-to-market engineers we have those since a couple of months one of them is here patty they have a business background but also they're very data-driven they're very focused on technology and you need to have that mix and then there is the business and the business in our case for go-to-market is marketing sales customer success or so your entire post sales motion and it's important that you bring them all together we started this ai-powered go-to-market working group, which has about like 15 people, and it keeps growing.

6:45Because we have seen cases where our data systems team built things with LLMs, but it was lacking the business context. And therefore, the models didn't work very well. Or we had salespeople wanted to do something, but originally they did not have the support from either data or systems or rev ops. And you need that as well. You need really that combination. We also made the initial working group fairly large, 15 people is a lot. And we did that also to have like broad coverage of all the functions and to make sure that we, as we start to change our culture to become more AI powered, that we have more people participating in shaping the direction.

7:32Then we said, okay, we started this the AI-powered working group. We had the Slack channel and then we had some initial ideas of like where to start. We looked at some use cases where we thought we could have a lot of impact. And then what happened, we started working on a couple of these use cases and then the ideas started flowing, including from myself. And we saw more and more opportunities and then people started to share those opportunities in Slack channels, people were raising their hands. and the problem was that as people started then to work on these new ideas that we hadn't finished the first one so at one point it started to spiral a little bit out of control and we said okay we probably need to have to have a better framework to prioritize really what we're going to work on because if you want to go from experimentation to skill and you want to drive more transformation then it's going to be more work to make that happen and therefore you really need to prioritize us.

8:34So we took two simple frameworks. One is jobs to be done. When I was in Dropbox in the past, our product teams were always talking about jobs to be done. It's a concept from Clay Christensen, who wrote about the jobs to be done from a product. He had this whole story about the jobs to be done of a milkshake. But we are applying it to our go-to-market roles. and the jobs to be done is what is this role hired to do for? And we look at all the different roles in go-to-market. So SDR, sales, customer success, solution engineers, what is really their job to be done? And what our go-to-market engineer will do then, one of them shadowed an account manager or a number of account managers for two weeks.

9:24and she found that the account managers were spending their time in seven or eight different systems they in order to do to perform a simple task they had to switch from one system to another bring together information and became very complex and they started doing readouts of these jobs and say hey this account manager job is losing two and a half hours per day here they're spending three hours a day there. Per week, they're working in all these different systems and we started mapping these different jobs. Now, the next thing that then happened is we saw like there were a lot of these examples in all our different roles.

10:05So then the question became, now we got all these different jobs. How does it all now fit together? And we said, okay, simple framework number two, let's look at our customer journey and let's see how these different jobs to be done now fit in. And so we started mapping them to our customer journey and we found out, okay, we have some for the SCRs. We'll talk about intent. We'll talk about win-loss and so on. And this was important to give our teams context and to show how these different things fit together. And I think it's important to look at like, where do you have a problem in your growth?

10:43Where do you have a problem in your business? that you want to solve and how do you prioritize where you start working with AI. The next one then was you need to build a culture of AI in go-to-market. Transformation is also very much about like are people adopting it? Are they really embracing it? We spoke about like what are the characteristics that are that are very important to build an AI powered go to market. For me, the number one characteristic is curiosity. We see that the people who are very curious, who are really leaning in, trying to figure out how this new world, how it's working and how it's evolving very, very fast.

11:30They are the ones that help you drive things forward. But we felt it was very important to build this culture of AI in go-to-market. And this AI search week was for us very important. But when you look at culture change, we adopted, or I adopted always this formula, which is the effect of a plan or of a transformation, in this case to AI, is the quality of the plan times the acceptance. And so five times five is a lot bigger than 10 times one. So the acceptance part and the creating of buy-in and getting people comfortable with how jobs are changing, how their work is changing, is a very important part.

12:19And that journey will definitely continue as we continue to make progress. And so what do we do? We try to make AI a habit by leading it, by showing it, and by celebrating it. let's talk about the first one the deleting it we use gong is one of our platforms gong also uses a lot of ai i'll give you an example of the role modeling when i do like the deal reviews and we bring in account executives to talk about their deals every once in a while one came up with like big power points of like how they're going to win this deal. And I would always go like, okay, please go to Gong, open up your account.

13:08There's new functionality in there. There's this little AI sign going there. Now we look for the account brief and everything is there. And in real time, we would do away with PowerPoint. And then the next time you do these reviews, the reps like really learn it. And I think it's really important that my leads do this as well, that they really look at the tools, get into the tools, and teach the team how to use the AI to their benefit. There's many applications to do that. There's also a great example of handles. Your AI can help you with handles as well. Second one, sharing and inspiring. We try to put a lot of our own teams on stage and have them share what they have built.

13:53So one of the teams built an assistant to personalize customer decks. So a very simple assistant. There's one team that built like an assistant to answer RFPs. We let the teams share those stories and inspire their piece and the rest of the organization. And the final thing is then also celebrating. We reward people who really help us drive this AI-powered go-to-market. One of the things that we did is we have every year our President's Club. Early in the year, we announced that we actually have two or three seats this year for the coming President's Club for the best contribution of AI. We're going to nominate a number of projects, and there will be real seats in that President's Club.

14:44Next year, there will be even more seats. Then, lesson number five, great AI comes from your stack, but then equally important, your context. And we heard a lot about that this morning as well. When you look at our stack, initially we take the view, let's not go out and buy all these tools. Because usually the tools are not like the panacea. there's usually a lot of work that you need to do in your workflows, in your data, in other things. So we said, okay, let's start with what we have and start building from there and then add LLM to it and then go from there. And this is our initial stack. So we are on Salesforce.

15:32You can also be on HubSpot. We made a big bet on Gong earlier this year. and why did we do that? Because for a go-to-market organization, the customer conversation is obviously a very important source of data and we felt that conversational intelligence with Gong or with someone else was going to be at the heart of our AI strategy. We also rolled out Qualifieds but not initially for AI. We did it because we wanted to have fast meeting booking and we wanted to have a sense of who would visit our website for intent. Only more recently, maybe in the last six months, Qualified started to adding AI capabilities and we started using them.

16:22And I'll show you later how we do that. Snowflake, we already had. We put a lot more data in Snowflake, both structured and unstructured data. and then we overlaid our LMs with Amazon Bedrock so you can use like various LLMs. And we thought, okay, we probably also need to do a lot of work on our data. So we deduped all our Salesforce data because initially one third of our data in Salesforce were duplicates. So we had a lot. So we installed like automatic deduping. We also spent months cleaning our prospect database. We bought external data sources and also there we tried to clean them up, connect them to have also a great prospect database.

17:11And that is all work that you can do, which is sort of independent from your AI, but makes your AI much better. We decided to make that investment early on. Now what did we do then? We connected basically all our customer sources. On our customer conversations, we loaded sort of 5 ,000 gone calls in Snowflake. We added a lot of emails. We connected to Salesforce, brought that all together. And then the thing that was also important is that we added Personio-specific go-to-market knowledge. So we added our ICP definitions. We added our pitch stacks. We added our onboarding processes. We added product training materials and so on and so forth.

18:02And this is really critical to really train the LLMs and to make it specific for your go-to-market, for your customers and for your products. I also believe that it's not on one hand you need to have enough data there. But on the other hand, you also need to have relevant information. And one question that will come up is, at one point, this data will get still. So at one point, the AI models will get worse. So we're going to run into this question like, hey, how do we keep everything fresh? And how do we actually take all the data or data that are less relevant, how we're going to take that out?

18:43We haven't answered that question yet. So in the next update, we'll give you an update on that. And then, so those were initial learnings that we had on how to approach the transformation to AI. And now let's look really into like four use cases that we have. Use case number one was about everything about win-loss. So we wanted to have more AI-driven customer intelligence and understand more why we win deals and why we lose deals. And the problem was the following. Our reps would fill out Salesforce after they had won a deal or lost a deal. And when you look at them, a lot of data points were great, but there was always like 30 % was due to other.

19:37and even in the cases where we had good information from the webs, we always wanted to go deeper and figure out, okay, yeah, let's go deeper on like why this is happening. And we felt that the insights were simply not deep enough. We then said, okay, how can we leverage AI to do that, to go like to get more insights? And we went back to like all our conversation data, our emails our salesforce data again we loaded that in snowflake and we build then this gpt for go-to-market and the goal was to really better understand our loss reasons our win reasons and really also get competitive competitive insights and our marketing team had recently updated all the battle cards.

20:32One of my product marketing leads is here. They did a fantastic job. But we were asking like, okay, can the GPT now enrich these battle cards? And what we found was that we were able to add like maybe 10, 15 % to the battle cards that initially was not that clear to us and that's number one and secondly battle cards are typically pretty fast outdated so how can you get to like a more continuous process of updating your battle cards and we were able with with the llms or with the gpt to like have more dynamically updating of our competitive battle cards and we will keep doing this as we keep adding like customer calls to the to the gpt so that was that was the win-loss use case but you can also see that this this approach can really evolve to like other use cases as well and ultimately that gpt can evolve into like a go-to-market brain you could do rap coaching from this you can do marketing campaigns you can give better product feedback on that.

21:44So we have analyzed also where we see a lot of demand for like certain features. And we have now much more data-driven approach to go back to our product teams and say, based on like 10 ,000 calls, this is where we see this product is doing really well. And here do we have, here's where we have weaknesses where we can still build. And that data-driven approach is also gives you a lot more credibility to your product and technology partners. But again, this will be a journey to get to a broader go-to-market brain. But you can see that over time, you will be able to address more and more of these topics.

22:26And how this will evolve, exactly, I don't know. But I think when you lean in and when you start doing this, you will see more opportunities. Then let's talk about another assistance that we built. This is for our expansion SDR. Our expansion SDR does cross-sell new products into our existing customer base. We have about 15 ,000 customers, so we're looking to cross-sell into them. The problem here was that when we did our jobs to be done mapping, that we found that every expansion SDR was spending two hours a day finding customer information to really make a relevant call. So they were looking for account health that was in amplitude or some other system.

23:18They were looking at like which contract this customer had. They were looking at many different things and it took them two hours a day to really prepare the calls that we're going to do. And we said, okay, let's change that. And we built one of the team, one of the go-to-market engineers built like this system. And what you see is it's actually connected to Salesforce. So they go, the EZR goes into the Salesforce instance where they work in. They type in the name of an account and then the GPT start collecting information from like maybe 10 or 20 systems. So it goes into Snowflake, collects all the information, makes that in a format that is relevant to this use case, a cross-sell use case.

24:07And then it will also come with a recommendation of what to do with this account. And it will say, hey, this is a great account. This is a green one. But there could also be reasons why an account is yellow or red. And again, and then basically the ESDR can act on that. Now, what is now the, people always ask, what is the ROI of AI? But here you see, for example, that the research time that an ESDR spends on this work went from two hours a day to 15 minutes. Not only that, you see here the pipeline that an ESDR is adding with this assistant. And basically, the pipeline per FTE generated went up probably like 2x.

24:55And there's many of these where you start looking at all your processes, all your functions, and you go to market all your jobs. There are so many cases like this where a lot of time is being wasted on either bringing information together from different systems or even prioritizing it. We have actually, as I said, now 400 of these assistants. Now, the top 10 does probably 80 % of the value. But we keep coming and people keep coming also bottom up with new ones. Then let's look at intent. Here the question is, how can we make our outbound motion better? And the approach that we've been taking on generating pipeline, most people, unless you're in an AI company, are always looking for more pipeline.

25:48And so we went through this journey where we want to find the right accounts, the right people or the right personas, going out with the right message. The LLMs help you with always tailoring the right message, but then also the right time. And right accounts, I mentioned that we updated our prospect database, but we have also applied account scores to our accounts. And this is really in line with our ideal customer profile. So we score all our prospects ABCD. Then we say, okay, let's now go out to the right people with the right message. And to do that, a never-ending work is work on your persona and contact data that is still in progress.

26:33We rolled out Gong Engage. We had Groove. That didn't work very well, so we rolled out Gong Engage. But you can also use Sales Loft or Outreach or any of the others. And then we use our LLMs. But then the question is right time. and this is often often the very difficult question so which prospect is in a buying cycle and how do you know and here you can buy a tool but here we decided to to do that ourselves and our data science team built based on a number of signals in now also an intent score which is dynamic and it's based on like did someone come to our website is there a former user of a platform that went to another company.

27:18Is it a great G2 or Trustpilot score or any of those? And you can really continue to reach that. And again, we took the view that all these tools need to come together in the platforms where our people already work. So again, this is back to Salesforce. We blanked out the names here, but you see the account score. these accounts are all A that is a static score but next to that we have the dynamic intent score which is now refreshed every day depending on whether the signal is changing in the market and then we show these little flames so three flames, that's where you start and this is all together in the place where people already work.

28:08Now We launched this intent model a couple of months ago. Again, you need to fine-tune it. We saw that initially the model was not great. It was picking up some signals that we didn't think were good. We changed it, and then it got better and better.

28:27Last use case is AIChat or the AI SDR. Here, for this one, we used Qualified. and I'll talk more about it. And I do believe that you can pick one of the vendors out there, but the goal is to go like really, really deep with it. And I'll show you how this one works. This one is really when you go to our website, you can start chatting with the AI chat.

29:01And our chatbot is called Nia. And Nia starts chatting here. And we do say here that a prospect is chatting with an AI. So it says, hey, I'm Nia, your friendly AI chat concierge from Presodio. And then what we wanted to do is like the moment we get like a demo request or a hand tracer, we wanted to like quickly, quickly book a meeting. And the idea here is that your best leads, which are demo requests or can I talk to sales, that you apply like high velocity. We live in a real-time world. So waiting for a week or 10 days to get like a demo booked, that seems like very, very long in today's time.

29:50and you see here that the prospect started chatting with Nia, requested demo and then immediately a meeting is booked. You can say, okay, this is just a meeting booker. But what I think is what is interesting, first of all, when you look at the last seven days, 140 meetings were booked. There were like 200 ,000 sessions on a website. So a lot of you see like, okay, how many conversations are already happening? and we're probably a month, six weeks into this. The other thing is that what is very interesting when you start like reading the chats, you see all of a sudden you see that customers have questions about your product.