Eps.93 | The system speaks first: Personio's Amir Jirbandey on context, agents and the future of HR tech

Speaker A: The systems we're using are not fully context aware. So the what next aspect of it when it comes to identifying, you know, how do we break through into this audience or what's the right channel mix or strategy, campaign, messaging, whatever it is, the component that's missing right now to be able to do that using AI is the contextual awareness. They don't know the market necessarily as well as we do. They don't know the nuances of the ICP in terms of where they hang out, how do they like to buy and all the intricacies that goes around that particular, um, segment that you're looking to break into or they don't necessarily have access to everything that's been done before in the company.

Speaker B: Great. Amir, thank you for agreeing to join me on the Enterprise Thought Leadership podcast today. I always love speaking to marketers, senior marketers, because it's the realm I come from. So I'm sure we're going to have an interesting chat today, particularly with all the changes that are going on and it never seems to stop all the change within marketing and we're see acceleration generated by artificial intelligence. Maybe Amir, you could just introduce yourself, let us know your background, where you work and we'll go from there.

Speaker A: Thanks for having me, Tim. It's a pleasure to have this conversation with you. As you already said, my name is Amir. I look after uh, international marketing for Personio. I've been doing B2B tech marketing roles, commercial roles for the last maybe 15 or so years, uh, a little tad longer. Essentially I'm a startup guy and international for us in personio is essentially all European territories outside of Dach. But this year we have our predominant focus on Benelux, uki, Nordic, Spain and a couple of other territories. So it's more of a kind of expansion role.

Speaker B: And when you say a startup guy, so what are the startups you've involved in? What are the sort of journeys you've been on?

Speaker A: The first startup I joined was a company called mailjet. It was a French startup, uh, based in Paris. I was looking after the UK expansion at the time. I got a taste for the VC world. Not that it means it could be frivolous with money, but just the uh, agile pace and quick potential growth that comes with it. Um, and since then I predominantly stayed within the startup world between kind of series A to series D, uh, usually during expansion phase, either going up market or expanding to a new territory. Some of the other kind of notable organizations I've worked at were Paddle, Treatwell and now Personio, there's been a couple of stints with smaller startups, one Australian one and uh, my previous role was at a company called Papercup where I was the first commercial hire, um, pre revenue seed round. They had this AI innovation which they didn't know how to take the market. So really going back to the fundamentals and that was a fun ride until we got acquired in 2025. Before I joined Personio, uh, it's interesting, isn't it?

Speaker B: I think there's two sides to the startup and scale up, isn't there? Where you've got this VC money and so you have this kind of Runway or you have a huge growth appetite and there's usually funds available to be invested in tech or in people to grow faster, uh, et cetera and invest in advertising, all the rest of it I guess on the other side you've got this kind of pressure, haven't you, around results and reporting at uh, specific milestones like quarterly reporting and I guess it could be probably quite stressful, can't it? I mean it's quite intense. I've got a lot of friends actually have worked in these sorts of organizations. Is it just startups and scale ups you've had exposure to or have you had other work experiences with organizations that don't have this sort of same Runway and pressure out of interest?

Speaker A: Yeah, I think my first foyer into marketing was as a PE backed systems integrator. Uh, we were working with some of the kind of latest innovative vendors in cyber security, routing, switching, amber row assistant integrator in terms of providing those hardware and services to the financial industry. With the PE backed I was relatively junior then so I didn't necessarily have the pressures as I do now. But the main difference between them, they had a very clear end goal in terms of this is what we're going to do for the next two and a half years, three years and the end goal is to get acquired or acquire someone else. So there was a very clear exit strategy where in the startup world, depending on the kind of the profile of the founders and the journey they've gone through at the state of the markets, sometimes that's a little bit more ambiguous. So you don't necessarily have that light at the end of the tunnel. Um, I've had a very brief stint at a public company in Australia. It was actually really popular for quite some time for startups instead of raising VC funds to use SPAC to go public and that's how they raised their funds. So you don't list yourself, you get acquired for $1 by another company that's already listed and then that's how you become listed. So SPAC is an acronym, uh, for that type of mechanism to go public. And we were operating like a startup, but you had all the restrictions of a public company in terms of your finances. What you can say, what you can't say is a lot more different compared to being privately held. So those are the two kind of call outs in terms of having different types of pressure, should we say? So I don't think you ever relinquish it. It's just a slightly different types of pressure and slightly different parameters that you've got to operate in.

Speaker B: Yeah, sure. And so the sorts of marketing initiatives and marketing strategies and campaigns that sort of, you've been familiar with over the not too distant past, what are we talking about? Are you specializing in one particular area or do you go across the board like multi channel? Where does your experience like.

Speaker A: Yeah, that's a fun question. For me. I'm a jack of all trades and definitely master of none. Um, it's potentially how I fell into doing marketing to begin with. I was a computer science graduate, not really interested into the developer and coding side of things, which I regret now because I could have been a lot richer. And at the end of it I was scratching my head in terms of what I want to do next. And I had this really amazing professor who was a, uh, kind of UI UX professional working in the field at the time when he was teaching us. And it turned on a few light bulbs over my head around the possibility of all the elements that go around E commerce, basically the psychology behind how humans interact with websites and technology. I knew there was something there, but I didn't know what it translated to. And after doing one year of IT recruitment when I graduated I realized sales is not for me. And at that systems integrator I managed to carve myself a marketing role. So because I fell into it, I ended up being a jack of all trades. It was the first marketing function at that company. So there was no one to learn from. So it was just trying to identify what I'm good at, what I'm not, and then just learning some of the fundamentals of marketing. And then from there onwards being a generalist kind of suited me quite well. Working at these different startups, most of them focused around growth and demand generation. So uh, that's where most of my skill sets lie now. But essentially just identifying some of those key marketing foundational elements from position and messaging, targeting, segmentation, and then moving upwards in terms of what kind of channel mix and what levers we can pull for growth. So again, it's worked out well for me. And we had this guest speaker from OpenAI, she does B2B marketing for them, leading their EMEA function. Um, and we asked her the question, who's at risk the most from a front marketing function when it comes to AI taking over your jobs type of scenario. And luckily for me, being a generalist. Being a generalist was, was at the bottom of the list. Which was lucky because again, I'm not good at any one thing. I'm good at putting the pieces of the puzzle together and work with professionals who are much better at executing it than I, um, am.

Speaker B: Yeah, let's talk about that. That's really interesting. So I mean, first of all, that's quite a challenging question to ask or ask someone at OpenAI. But what was at the top of the list? Do you remember what she said? Was it he or she, she. What did she say was at the highest risk?

Speaker A: To be honest, I don't remember. But I remember something hovering around the top three was just essentially most highly specialist roles within marketing. You can take product marketing to design to performance marketing, et cetera, et cetera, where we're already seeing some of the different use cases from Claude Enterprise, et cetera, et cetera. We pin in for now. It's for good, uh, in the future. Mm, let's see what happens. But mainly when you have those specialist roles that you can train one particular model on those specialist skills to repeat itself. Yeah, um, I think that's generally what's at risk the most. But this is like a five year vision, not tomorrow.

Speaker B: Yeah, I know, but it's, it makes a lot of sense. How do you see when we talk about the impact on jobs or people or organizations or team structures. Go wherever you want, Amir. But what do you see as what's going to be changing over the next year or two? And how should marketing leaders be thinking about it?

Speaker A: This is the, uh, question we were asked quite often, especially in my previous role, because we were a native AI organization and this is pre chatgpt era. So it was before AI was as cool as it is now, or uh, how popular it is now. And our answer was always the same and I think it still holds true. But we're essentially working in AI voice arena, selling into major studios in Hollywood and large publishers on the East Coast. And when we were faced with that question, the answer was always that AI is a tool that's Going to enable you to do 20, 30, 40% of the more menial work that you do. The bottom of the rung work that you do help you to do those faster, better, more efficiently. So that way you can focus your kind of craft on the top end of the spectrum. So if you take a movie for instance, and we're looking at AI dubbing or any form of AI voice, you're still going to need those very prominent protagonist actors and actresses. However, where AI may have come in into that scenario is you have secondary tertiary actors, all those extras, et cetera, et cetera, who may have a line or they may not. When it came to localizing those voices, instead of spending money in terms of human localization, you would do it through AI, but still spend that money on that human localization when it comes to the main actors. Because those nuances in language, uh, is an art, right? And as we know, that's one of the hardest things to imitate, if at all should be imitated. It's very similar to what we do in marketing. Marketing. Again as you said, as you mentioned, in terms of the software engineer case study for bringing it to a marketing level, some of my colleagues in performance marketing, for instance, now they're spending a lot less time creating those assets and writing briefs and creating monthly reports, et cetera, et cetera. And they've upskilled themselves very quickly to be able to have more strategic conversations around dynamic ICP and segmentation and how this fits as kind of part of a wider campaign or strategy or tactic or whatever it may be. So essentially they are becoming more of a360 marketeer, but over indexing on one particular channel, one particular field, and to be at be part of conversations which they potentially wouldn't have happened. Would have happened, yeah, two, three years ago.

Speaker B: I think this is a really interesting area, isn't it, for leaders like you and many others, which is your team, once they become fluent with AI and you've built the automations and the systems and they're working, right, it's like what next? Ambitious organizations want to grow, right? They want to look at uh, innovative ways that they can capture market share, grow faster, be the competition, et cetera. But you still have to figure out how do you do that, right? So if you give your team 50% of their time back, which is easily what AI can give you once you've got it set up properly, right. Potentially more. It's almost back into that kind of brainstorming, isn't it? Get together for an offside or Whether get the whiteboard out, what can we do next? And is it about. Oh, actually we don't have a channel partner, uh, division who wants to step up? Okay, me. Great. All right. Who are all the companies we want to reach out to? Boom, boom. Right. So you've all of a sudden you've got time to actually go and build these new kind of systems and these new relationships, et cetera. Would you agree that that's the sort of the mindset, would you say?

Speaker A: Partially. I think maybe that's where majority of the market is that adopting AI into their workforce right now. However, I think the underlining message as part of the scenario you're describing is the systems we're using are not fully context aware. So the what next aspect of it when it comes to identifying, you know, how do we break through into this audience or what's the right channel mix or strategy, campaign messaging, whatever it is, the component that's missing right now to be able to do that using the, is the contextual awareness. They don't know the market necessarily as well as we do. They don't know the nuances of the ICP in terms of where they hang out, how do they like to buy, and all the intricacies that goes around that particular, um, segment they're looking to break into. Or they don't necessarily have access to everything that's been done before in the company, your knowledge base, your playbooks, et cetera, et cetera. So one thing that I'm very excited by is especially at personio us, uh, now having the tools that are able to become context aware, um, again, still not 100%, nowhere near 100%, but it's a lot better than it used to be, especially using Claude Enterprise. All my chats are, uh, now context aware. I've created this co work agent, I call it my cmo, who I, it's my sparring partner. I went back onto it this morning just to have a conversation about this podcast to give me a few nuggets from, um, some of the conversations you and I had before Tim. And he asked me, paige, do you want to continue this conversation as it was in, In. In. In Karl, or do you want me to just summarize it and then we can start a new conversation? My point is it's completely aware of everything else that I've fed it historically without me having to do anything. And this is the most basic setup that now you have. And what we do in operation here is just again, through MCPS and all the different connectors make sure. That it can read everything. So as part of that same conversation I was like, hey, can you look into this other Slack channel where we talk about our, uh, AI use cases? And we had this hackathon. Can you pull out those couple of stats and bring it forward here? Again, I didn't need to make it aware of those conversations because it has access, it becomes context aware. So the analogy that you described earlier, I think it's going to, it's going to be a thing of the past pretty quickly with more context aware systems.

Speaker B: Yeah. And a thing of the future in terms of humans getting together, figuring out what's next and being creative and thinking strategically. Right. And then taking on new previously unimagined tasks and opportunities. But what you say is, I'm great. I'm really pleased you brought that up actually, because there are different layers here, isn't there? There's agentic AI where you're talking about Claude code, obviously Claude cowork that sits on top of Claude code. You have other uh, gentic platforms that you've got codex from OpenAI, you've got goose, you've got, I mean there's OpenCloud. These are the harnesses that these models can use to be able to access multiple systems simultaneously and provide insights that the main brain can look at and take a decision on what to do next. And even if you've got the tools and you've got people who know how to use them and understand, even at a, uh, relatively simplistic level level, you've still got this issue around data, uh, relevance and memory and the ability to make sure that information that resides in all of these systems is not stale. Because if these agents are serving you up information that kind of is been and gone. Right. It's like all of a sudden the context that you're talking about is polluted with stale information. So I think. Are you involved on the AI Council at Commons? Keep frame for the audience who personio is in a second. But you mentioned, I think you were on the AI Council, aren't you?

Speaker A: Or you're involved the GTM one of it. Yeah. We have different pockets of AI integration and use cases. The GTM one again, it's nothing fancy, it's just a Slack channel where it was initiated by our CRO, who's very forward looking in this regard. And we started to brainstorm ideas in terms of what we can automate, what we could do better, et cetera, et cetera, and place a few big bets and a few small bets individually. Uh, so I'VE been more of an observer than a player.

Speaker B: But, uh, yeah, but on the topic of what we're talking about now, it's like that sort of technology layer is relatively easy, right? It's the data. The data that has to be. That has to be. Right. In fact, we were talking to a client recently about so many conversations that are so rich between colleagues or on WhatsApp. Right. Do you know what I mean? Or WhatsApp calls. Okay. And whilst this is very easy for us humans, you just pick it up because it's the sort of device and the application that you're used to really, for the future of work. It's criminal. Right. Unless of course, uh, it's a business account that agents can access. Right. Because I think the whole game here is about giving AI agents access to as much information as possible and then ensuring that information is not stale, as we were saying before, at least have

Speaker A: the ability to identify when it's stale or not. So if you can feed it again, you just go back to simple machine learning concepts, which is, here is what good looks like. Go, uh, look for good. And here is what bad looks like. Go look for bad. Or if you're trying to get, uh, kind of a computer vision model to identify a chair, we just got to feed it loads of different pictures of chair and things that are not chair that look like chair like a table.

Speaker B: Yeah.

Speaker A: And then over time through, through the kind of neural network that it creates, it ends up identifying chairs at a very high probability. Again, it's never 100%. It's similar concept now. And it's funny that you mentioned the core aspect of it. When I was working for that Australian startup that was on a spac, which I mentioned earlier. And by the way, I quickly googled spac. It stands for special Purpose Acquisition Company. We were doing dubbing. Sorry, not dubbing. It was called dubber, not dubbing. And uh, we were doing call recording in the cloud at the time. They were like the second or third highest user of AWS in Australia. So they were relatively early in this area. And this was a really old stagnant industry because everything was done on premise in hardware because of data sovereignty. Essentially. Some of the early models that we were able to get on AWS, and this is going back almost 10 years I think we were able to do basic sentiment analysis and different types of insights that you get from your calls. Fast forward into today. Now we have beautiful things like GONG and many other players that do something very similar and so much more. Um, so, yeah, we do feed all of that into some of the different other use cases, which we'll talk about later, I'm sure into our AI internally. But just showing up what good looks like and what doesn't, I think is still really important. Going back to my computer science and uh, this is older than my course, but when it came to kind of first class of programming, building databases or anything like that, the first thing your lecturer would tell you is garbage in, garbage out. Um, so having the right data, as you said, is paramount to be able to get the right insights.

Speaker B: Yeah. All right, let's, as you said, we'll get into some use cases a bit later and it would be good to put the data lens on those when we get to them. So let's talk about Bosonio. Maybe you could just give us an overview of the business, what you do, how you're positioned in the market. Appreciate your. I think you're responsible for all marketing in Europe outside of dac, is that right?

Speaker A: That's correct, yeah. I wouldn't say I'm wholly responsible. We have a multi layered team, but essentially the overall pipeline number is on my head from a marketing source perspective. So yeah, essentially my role is focused on marketing source accepted pipeline. That's essentially the North Star. However, our teams are quite devolved so we have central teams when it comes to rolling out individual kind of channel execution from performance to web digital experiences, content, et cetera, et cetera. My role is essentially working with kind uh, of VP of International and other senior leaders to set out our plans for the subsequent year and quarters and half of the year. Um, and then work with all the different individual teams as well as my own team who are basically the regional kind of leads for each country or region to execute that plan. But then make sure that we're playing a partnership role with anyone in a central function to make sure they're executing at a level that we need in that territory. And uh, we hold each other accountable. And as you could imagine, it's highly collaborative with our sales function as well.

Speaker B: Yeah. All right, let's talk about the product. So an HR is. It is HR Capital Management System. Is that what it's called?

Speaker A: Hr?

Speaker B: What's the case?

Speaker A: Yeah, just HR tech. Yeah, or hris. It's an all in one solution for people in HR leaders. So the core components of any kind of HR tech is being able to have records of your employees and then everything that goes with it in terms of time, tracking, ship planning, leave, onboarding, upboarding, paternity leave, et CETERA et cetera. Um, and then there is additional add ons where you want to have a single source of truth. So for instance, if you have your recruitment elements attached to it, then you can have an end to end life cycle from hiring all the way to exit. There is performance and development, there is learning modules, surveys, whistleblowing, the list goes on. Then there's the marketplace where again you have multiple integrations such as payroll and many other things that you could think of. Um, we were founded about 10 years ago in Munich. Munich, Germany. So we're a German company, I like to say German engineered organization. I haven't got the sign off to use that in our marketing yet. But I think he has really positive connotations and we tend to serve mid market. So uh, usually organizations of around 102,000 headcount. That's our sweet spot. Yeah, that's it in a nutshell. I won't give you the whole elevator.

Speaker B: I mean it's a, it's a busy market, right. And it's, I mean there's a lot of big players appreciate that. Probably in the enterprise, your SAP workday and others. What is it? If I was to say give me three key differentiators or USPs of your platform, what would you say?

Speaker A: We had a massive rebuild from the ground up on the platform which coincided with a rebrand last year. But if you look at some of those players that you mentioned, they're ridiculously old, right? Some of them older than me and their pace of innovation has been relatively slow because historically this has been a very sticky market or a sticky solution no longer unfortunately. But some of those larger organizations have been using the same system forever. Um, so sometimes they just don't know what good looks like now with what's available and how far innovation has come. Uh, that's one number two, I think one thing that we started to over index on based on uh, a lot of feedback from our customers is level of customization, especially when it comes to reporting. The HR and people job title or function is the fastest growing in Europe and um, probably even faster in, in the uk. The reason behind that is, the reason behind that is, is because they're doing more jobs than they used to, they're wearing more hats than ever before. So it's not necessarily that you need more HR people as a kind of a net gain on top of your organization, but it's because they're replacing other roles that you may have had. So for instance, if you had maybe like a very junior legal person that focused on employee law or labor law. Now that's kind of expected for the kind of HR leader or the people leader to do. If you had certain counselors within the worker, uh, work environment or in the office now is expected to be the HR leader when it comes to kind of strategic decisions when it comes to growth of the organization. That's something that might have sat with the operations team but now it sits with the HR leaders. So they're wearing more hats than they used to and their role is becoming even more the word. I'm looking for a grander word than valuable but it's paramount to companies existence and growth. So it's growing and that's why we're still seeing a good level of growth in the market.

Speaker B: And um, what's the AI story? Appreciate that sort of headless software is, is where it's going, isn't it? And there's the SAS Copalypse is causing concern in the markets, etc. What can you say about how the HR tech sector, uh, how is that, how does that fit with the current AI narrative?

Speaker A: It's probably twofold. One of them is AI to be used by those HR and people leaders. But then there is the uh, probably a bigger uh, subject to talk about which we often do with a lot of our community events and content that we write based on the feedback from the market and that's the workforce shift because of AI. How do we manage that? We've talked about that 20, 30% in terms of uh, efficiencies gained by certain teams. But that's us sitting in a privileged position to say that there are definitely a lot of roles which are going to be negatively impacted by AI either directly or just as an excuse. We've seen the hyperscalers shedding a lot of headcounts using AI as an excuse. But I just personally think it was the Overhire from the COVID era. And how do you manage that? Um, how do you manage the future changes that are coming? Because innovation is happening so fast and people using AI in so many different ways. So that's the second topic which I'll um, put a pin in. But that's the bigger topic, the immediate topic that's more related to Persona and how we operate is how do we enable them through the right tooling. One element I alluded to earlier as part of our usb which is the level of customization and reporting. So one thing that we launched last year, which is relatively at a fast pace improving over time, is reactive and proactive insights. So through The AI assistant that we have within our system through Reactive Insights, you can have a very kind of natural language conversation with the assistant to pull big chunky data that you would have had to download as a CSV or an Excel and dig into it to come up with, I don't know, a pivot table to see X, Y, Z. But now you get that in an instant. Through the instant. That's the Reactive insights, which is potentially a bit more of a commodity. But we talked about garbage in, garbage out earlier in terms of having the right data. We're in a privileged position as an all in one solution to be a single source of truth for everything people. So we are able to provide higher level of insights when it comes to that reactive element. The fun part is the proactive one. The proactive one is not just for people and HR leaders, but any people managers in the business where you get reactive, proactive insights coming to you saying, hey, did you know that your team hasn't taken any time off for the last two months? Here is a nudge to it. It sounds simple, but it goes a long way towards making sure you have a happy, healthy and growing workforce. So that's the thing I'm more excited about.

Speaker B: Does that come how. Obviously, um, you've got to configure that. Right. Depending on what the information is that you want to push out. Is that. So how does it work for your customers then who are using personio? Is it quite technical doing that or is it just a case of, I don't know, talking to the agent and asking it to create a. We're often known as cron jobs. Right. How does it work for the listeners who might be interested in your platform?

Speaker A: Uh, there's a lot of good functionality right at the box. So one of the things that we pride ourselves in initially when customers on board is we have a very highly skilled and dedicated implementation team. So it's not like you purchase it and you go and figure it out. We go through the journey with you to make sure you're fully onboarded and comfortable. So yeah, a whole bunch of it comes straight out of the box. But anything subsequent could potentially be programmed at the point of implementation. Again, just the asterisks at the moment is not fully rolled out to everyone. Customers have it. Some others are ah, getting onboarded now. So hopefully. But by the end of this year it will be uh, across everyone.

Speaker B: Yeah. What about performance management and I guess, yeah, being able to use technology to analyze how individuals are performing against their KPIs.

Speaker A: So that's not something that I'm aware that we've uh, given to the whims of AI yet the performance management side of our system is predominantly around how do you do your annual evaluations or biannual evaluations that you do the 360 evaluations in terms of do a self report, your manager does report, you get peer reports, et cetera, et cetera and then it consolidates that into a certain type of report to see if you're on track, off track. So it's still very human LED aspect of it and not given to AI. We don't necessarily tie performances or okrs into that performance piece on Persona.

Speaker B: Do you think that's the future where you can literally if everything, if the intelligence layer can see emails, see Slack messages, teams messages, transcripts from teams, calls, Slack, whatever, zoom, you can actually tell how staff are performing. Right. To actually map filter that in to sort of look at almost like a gap analysis because obviously there are the whole things around employees might feel got the kind of. The Big Brother organizations are going to realize that pretty much all data now the game here and we might go back to what we talked about before about data accessibility, etc. But if you have got all of those touch points you can now program these AIs to be able to perform that role. The gaps from a managerial perspective or performance or a compensation perspective or whatever it might be, but almost like a kind of real time coaching.

Speaker A: Right.

Speaker B: AI now knows what good looks like because of all the training it's had and it can obviously be trained further and tweaked, et cetera. But to me there's a massive opportunity within hr. Right. In terms of how you can fully leverage AI if you've got your data in place.

Speaker A: Yeah, potentially to be honest it's something I haven't personally given that much thought towards and as you were describing it, it definitely from a personal level and makes me feel uh, uneasy just because of the Big Brother elements of it that you mentioned. I think we saw some announcements, I think it was meta not too long ago mentioned that now they're going to tap into a whole host of different types of comms within the business. They said it's for training purposes and nothing else. But you can already see it creeping in with some of these larger organizations. How it's going to be used, I think we have to wait and see. But I'm more of a Luddite when it comes to this subject. I think there's going to be a little bit of a Revolt potentially if anyone found out that their performance is going to be judged by AI analysis. And we saw probably glimpses of this at the beginning of the Industrial Revolution. So when it comes to anything that's to do with manufacturing or anything that's highly transactional from that perspective, I uh, think we've had this type of thinking at least if not the systems, but with more nuanced roles. I find it difficult to see this in a positive way impacting certain departments or certain roles.

Speaker B: Yeah, no, I think you make a really valid point and uh, it's got to be done properly. But I think if you look at it from the other perspective, the AI can actually help the individual. So if you're able to, you've talked about your clause co work, right? Where you tell it where you need help and it goes and looks at information, past information. Imagine that from an employee perspective, right? Where it's just okay, this is what I need to be doing. What should I be doing next? Or how could I have done something differently, right? I mean I'm not going to give you loads of use cases because I'm interviewing you, not the other way around. But what I have done in the past is I've actually run a few times run transcripts of whether uh, it be sales calls or client calls or whatever through an AI, a purpose built pull project to get feedback. And I mean it's very powerful. You should have said this, not that. Why did you say this? Why didn't you say, I mean it's incredibly powerful. And then if you take that simple example and then you overlay it onto an organization, these are the sorts of HR systems of the future. I think.

Speaker A: Yeah, I think that the analogy provided makes a lot more sense in individual pockets being able to utilize it in this way. Let's use sales coaching. You know we've been using GONG calls which uh, then flag certain um. Because you can't monitor every call, right? You're not going to sit down and review every single call a particular BDR makes it. Especially if they're doing, I don't know, 20, 30 calls, meaningful calls a day. Um, but what you can do is you can program it on certain phrases, certain type of sentiments and even match it to certain types of accounts or certain types of Personas that you're speaking with to highlight, hey, how could I have done that better? And our sales team are already using this as a self serve mechanism to improve themselves. So to this level, yes, I do agree with small pockets and small use cases. But one the Part that makes me feel uneasy and also I don't think it's going to catch on that quickly is the catch all that, that kind of the God view of everything that you're doing to provide you with uh, uh, certain type of productive ticks, tips and tricks on how to do your job well, mainly because I don't think it's going to be that ever that context aware and why it's never going to be that context aware is because us humans have to feed it that data and we're terrible at that when it comes to structuring data and feeding it. So until AI systems could go and create their own data that they need to be fed with.

Speaker B: Yeah, but surely if you have a policy as a company like all calls are recorded so you've got transcripts of every call, internal and external, you've got every slack message, you have a policy which is if you're going to be using WhatsApp, you've got to use it on the company. And it's just that's the policy. If you want to do anything personal, you need to do it outside of the work systems on the basis that the intelligence layer is going to be overseeing everything and helping fine tune the business. I mean I know that's what business owners and CEOs want to do. I mean it makes complete sense. Right. But raised a very valid point which is a change management consideration which is unless this is done very carefully and intelligently, it's not going to rub. Right. Particularly in the early days as people get used to it. Cool. Okay, so just in the last few minutes then let's just hit on some of your use cases right now within the business. You were talking, I think quite interested with the sort of the AI angle you're talking about gong as I know that you had, I think you'd brought in Snowflake and you'd loaded, I think it was 5,000 gong calls. I mean this is super cool. And then you're able to take obviously all the intelligence from those and I guess do stuff with it with the SDR team or the marketing team, maybe Amir, you could list a few. Happy for you to respond to that one, uh, in point in case. But do you have any specific use cases that you could share that would be quite interesting for our listeners in terms of how leading the kind of the European marketing initiatives etc. What you're doing that you're finding is working really well today, May 2026?

Speaker A: Yeah, absolutely. As I mentioned, the 111 drive that especially the GTM Council has been a part of or have been trying to push forward is just making our systems very context aware. So go as we touched on is a great source. So as you shared the number we've loaded it with over 5,000 calls and growing uh, but then making it context aware by marrying it to our own database which we first had to clean up Qu so I think there was something like 33% duplication which we got rid of before we fed that into our machine learning algorithm. And the machine learning algorithm that we were building was for us to basically target the right accounts. In an ideal world, every marketeer's dream or a salesperson's dream is to have the most accurate books of business where you know you're going to have a high propensity of winning. So uh, our aim m was to create a machine learning system just to do that. So when it came to building our books of business where sales and marketing on um, basically targeting the same people also keep on the same heat, same sheet, we need to make sure we have a high level of accuracy. So after we got our house in order, got rid of those degroops, using probably three or four different vendors for third party data enrichment, using every first party data kind of angle that we had ourselves, we managed to build a system to give us a kind uh, of a uh, tiered level. So we had an ICP and Persona that's our books of business in terms of where territories lie and those are tiered ABCD accounts based on propensity of winning and winning. I should clarify for us at the moment is accepted pipeline not closed one and we managed to see some drastic improvements. So if our base layer of our books of business we had I don't know, 30, 35% chance of turning that to accepted pipeline. In certain territories and certain books of business that went up to 70%. Um, and it was essentially as part of that big driver making it context aware but adding very kind of clear parameters in terms of what good looks like and what doesn't. Next stop is how do we now take it all the way to closed one. So that's the next phase that we're working on.

Speaker B: And uh, what sort of what things have you got spinning to make that a reality and high propensity of success.

Speaker A: So it's literally just building additional data sets to feed into the machine. Uh obviously the machine learning algorithm is always being tinkered with different signals that we receive, it's never static but just those data sets are probably the most important Part, definitely.

Speaker B: And you'd mentioned you'd recently bought Claude Enterprise, which is fabulous. What do you see? Because I think early on in the interview we were talking about, now you've got this kind of opportunity to do a lot more with Regentic AI. What are some of the use cases that you're thinking about?

Speaker A: That's a relatively wide question. So we started with our uh, kind of internal AI drive last year. We ran this kind of a, uh, month long program which ended on a week's hackathon at the time. The tools that we had access to, I think Lambda had the biggest uptake and then slowly we started rolling out other tools and systems. Fast forwarding to kind of Q1, end of Q1 we had a second uh, wave of that. And I think throughout that process and a run up to it, we ended up building like 400 different agents across the business. Some of them doing very menial, boring stuff, some of them doing funkier stuff. Uh, not all of them necessarily gtm. So for instance, I think one of the highest used agents was within our procurement team because we only have a team of two in terms of procurement. Um, they're doing a million different jobs through this Agentix systems that they've built, which has been fantastic to see. Another one that I use on a regular basis, system called Panda. So he aggregates a lot of our data, uh, again using Snowflake as our data lake and then being able to pull that out into CLAUDE directly. We have multiple systems for different data points, sales data points, marketing data points, et cetera, et cetera, using more traditional BI tools. Just don't cut it anymore. Um, so you know, the tablets of the world that we've been using for so long are just not agile enough. So we just built our own, uh, and that has had one of the biggest uptakes and it's probably one of the most accomplished systems that we built during that hackathon. But as I mentioned there's 400 of them, so there's quite a lot to go through. This was the one that I was really excited by.

Speaker B: And are you all encouraged? I think this is the last question for you. Are you all encouraged to experiment and build your own systems with the AI platforms or do you have a kind of a, Is it more of a centralized process where. So do you see what I'm saying? When perhaps not encouraged as much to do that? I'm interested.

Speaker A: Yeah, no, definitely, definitely the former. I think again, credits to Hannah and Philippe and the other leaders in the business as part of Exec and senior leadership team. They've had this top down drive in terms of creating the right initiatives, the right tools, the right space and enabling people leaders to help their teams, number one get excited by it, number two, dedicate time and space and number three is that thinking about real life use cases which they can use uh, these tools for. But then after that, after they created the space and provided all the rights elements for us to do this, they left it to our own devices. So now it definitely is an employee led initiative where we're creating things right, left and center, uh, and um, geeking out on it, casing it to colleagues and friends and whoever would listen in terms of big why I build. So it definitely has become an employee led initiative. We're actually hiring our uh, first head of AI and automation to be that kind of central person that becomes the glue to hold these things together because it can run wild. As you could imagine. Um, the only kind of restriction that we had so far uh, was on certain tools that we can use mainly because of data sovereignty as you could imagine. Because we want to make our systems context aware. You can't just feed that into any system. And we also want to make sure that we have really good integrations with all the different tools that we use. Again you can't just use any system for that. So that was the first limitations. But now how far CLAUDE has come and Claude Enterprise, it's become a massive unblocker in terms of the level of use cases that we're tackling today. Another role that just went live literally two days ago is our first kind of design engineer which is our ah, kind of AI first designer sitting with our uh, digital experiences team working on web projects. All these roles really exist. So I don't know how we're going to look for the right people but I'm really excited by it.

Speaker B: Yeah, a lot of curiosity and experimentation type people that have been, I guess have been playing with AI for the last year or two.

Speaker A: These are the people because job spec says five years minimum. Claude Enterprise Experience.

Speaker B: But yeah, yeah they only, yeah, yeah they only brought it out a few years ago. But because you did mention on our briefing call that that you actually sent a candidate away to go and get more clued up on AI. Because I think the future of hiring people is going to be you need to demonstrate that you can bring not only yourself but a huge amount of fluency with the various AI tools, otherwise we're not going to hire you at the moment.

Speaker A: Yeah, definitely. I think it was a version of that where, yes, uh, we're baking in some of these use cases which I mentioned on the back of those 400 agents, for instance, that we created. We just want to make sure that anyone that comes on board, to use Hannah's words very rightly, they're the ones that pull us to do more and better as opposed to coming to personio just to learn how to do these things. So we're looking for people to challenge us as they come into the business in terms of all the different use cases and applications of AI within the GTM kind of function or marketing functions. This is something that we bake in into our interview process more and more as we move forward. Uh, but we had someone that got stuck in the process between how fast we grab it basically. So when we started the process, we didn't have this kind of hard requirements, uh, to be self sufficient with some of the tools that we use. By the end of the process. That was definitely something that we were looking for. So we just asked them, hey, would you mind doing one more extra task? And I think just the nicest thing about that process was that person was actually really excited about the opportunity to show off their skills. They smashed it, smashed it out of the park. They'll be joining us in July.

Speaker B: Great. Amir, thank you very much. Really appreciated your time.

Speaker A: Absolute pleasure. Thank you so much, Tim.

Speaker B: So thanks so much for listening. If you're looking to transform how your teams work with AI, please visit cogniscale.com we upskill teams, give leaders the visibility to prove ROI and help you govern and scale. Get a free AI opportunity report by using the promo code podcast. Also visit TechPros IO to join over 4000 professionals who have participated in Challenge Forum, uh, roundtables, thought leadership interviews and industry reports. Please remember to subscribe and we look forward to catching you next time.