EP 88: AI agents now build the B2B shortlist | Otso Karvinen,

Speaker A: Welcome to Not Just Ads, the show where B2B brands discover how creator led content drives pipeline, builds trust and grows long term brand equity. Each episode we speak with business leaders, founders and creators who are defining their categories. I'm your host Marcus Sherwin and this is the creator economy for B2B. Hello and welcome today to Not Just Ads, the creator economy for B2B. Today I'm delighted to be joined by Otsu Caravanen. Um, Otsu Up M until recently led growth at Karu um is a uh, Finnish um B2B SaaS company. He since um recently went out on his own uh to launch a ah, GTM agency. Uh, am I right in saying agency or is ah uh, I guess so. Right.

Speaker B: Yeah, exactly. And guru as well is ah, uh an agency as well. So agency background, going into an agency.

Speaker A: Perfect. All right, well it's great having you on the show Atsu. Thanks a lot for joining us. I'm really looking forward to getting um, uh, your input on where you see the GTM industry going, where it's evolving and so on. So I'm looking forward to the conversation. Thanks for joining us.

Speaker B: Yeah, thanks for having me.

Speaker A: Great. So you've been in this space for about 15 years. Um, so um, now you've decided to go out on your own and um, how. I know this is very recent so um, maybe you could just share like um, a little bit around your background. Um, specifically in B2B just so the audience can kind of get a better understanding.

Speaker B: Yeah, yeah, absolutely. So I'm probably like very similar to a lot of marketers. Um, and I ended up in marketing basically by accident. So I was, I was studying in university and I was completely broke and I had to find a job and I was like what do I know? Like what skills do I have? Like what do I know? And I was always pretty good at writing, always pretty good at English specifically. So I became like a copywriter. That's my background. I started as a copywriter, worked as a freelancer for about five, six years doing. It was the early days of social media and especially social media advertisement. So I got in very early Google advertisement and Facebook advertisement and later on then LinkedIn advertisement. Uh, worked a lot with different companies around the world. A lot of US companies, a lot of European companies, a lot of them were B2B companies specifically. Um, and then I did a little bit of a detour into my actual field of study, which was hospitality. I worked in a couple of hospitality fields and then now uh, the past four and a half years I've been working at Guru, uh, which is um, uh, a B2B focused digital marketing analytics agency. So worked with a lot of different sized B2B companies from like in the low millions to the highest biggest companies were in the billions of turnover. So I saw very different scopes of uh, companies and how their go to market operates. So it was a really uh, very cool kind of learning path in the past four and a half years as well.

Speaker A: Nice. So when you say B2B analytics, what was the kind of the service proposition that you were providing those companies?

Speaker B: Uh, a lot of web analytics, digital marketing analytics around that, so around marketing analytics specifically. So data analysis and analytics and analyses around that especially Guru is very well known for web analytics in Finland and here.

Speaker A: Yeah. Okay. And then so I guess there's a big change now in terms of traffic, where that's coming from and so on with LLMs and how people are getting discovered.

Speaker B: Yeah, for sure. Yeah we saw that already like a couple of years ago. I think for the last year it was kicking off even more, even more. So um, I think it's in B2B it's, it's kind of like ah, a little bit also risky just to look at the numbers because a lot of the, what is influencing is happening. Like uh, I quite often use this visualization of an iceberg. So what you see at the top is the only, the very little but actually a lot of the way that people find information and what actually convinces them has changed from going uh, into web pages, reading about the companies on the web pages, uh, and then now using ChatGPT, using Claude, using different tools, using agents to do the research for them as well.

Speaker A: Yeah, yeah, yeah, yeah. No it's definitely, it's changed a lot. Uh, a stat I heard recently from LinkedIn is like 94% of B2B purchases now begin on an LLM of where they decide who's going to get on day one list. So it's uh, it's definitely changed a lot in terms of how, how people are doing that initial discovery as well.

Speaker B: Yeah, absolutely, absolutely. Even, even my own purchasing behavior like now for the, for the new uh, for the new venture, I've of course kind of research all kinds of tech stack, like what tools do we, where and how. And um, a lot of the decisions have been made because my agents have done the research and provided me with a kind of explanation of why this particular platform is actually and there's some great examples. Like there's an email service provider called Resend that has Grown like exponentially fast because all the LLMs are actually recommending it as an email platform rather than any of the kind of old incumbents in the market.

Speaker A: Yeah, yeah, it's, it's an interesting one. It's definitely. I think people that are on that um, are. Have. Have a. Have a. At the moment anyway. At least they have leverage and they have an advantage if you are getting recommended there. How. So talk to me a little bit about how the idea came up to go out and, and start your own thing. Was it something that was on your mind for a while and, and, and maybe you could share a little bit about the company then as well.

Speaker B: Yeah, yeah, absolutely. So I think realistically it's been all. It's always been there because I started my career as an entrepreneur. I was a freelancer. So I have this kind of entrepreneur gene I would say. I would say so it's always, always been there. I always have worked very uh, similar to entrepreneur, like in an entrepreneurial way. So I've been always thinking like at some point I want to do something like my own and I want to. And then the opportunities have kind of been there like a few of them being their discussions. But now um, I met my co founders. Co founders, uh, early on this year we started discussing like what, like what could we do actually and should we kind of put our brains together and our kind of thinking together and start something. And I think, and you probably agree Mark, is that the kind of how. AI uh, in GTM, um, in B2B marketing. It's really going like crazy now. The beginning of this year, like end of last year, it was still like people were still kind of you know, touching and feeling and what works and what doesn't. And then it just kind of exploded now. And I think the timing was one of the big things. Why I think, I think now is like a um M. Now is a window where this kind of move makes sense. And honestly I think also I think a lot of traditional agencies, a lot of traditional um, consultancy companies are going to struggle with the transition from the kind of old way of doing things and now into the new kind of agentic way of doing things. I think they're going to have a lot of kind of growing pains around that. I think it's very interesting roles in those companies, like change leadership roles. But that's not for me. I'm more of. A bit more of a builder. So I want to uh, kind of build from scratch if I have the possibility. Yeah.

Speaker A: So. And what can you share as well. Like what's the, the customers that you serve, the value proposition, what's the, what's the segment in the market that you're going after? And maybe you could share a little bit about your co founders as well.

Speaker B: Yeah, yeah, absolutely. So, so we basically uh, I think the easiest way to explain the company is an AI native go to market agency. So uh, we work with sales and marketing teams that especially um, have a problem with resources. So how we kind of put it right now and of course it's like a living document, our ICP and our value propositions and everything because we are a brand new company. But it's really about those companies that have high ambitions, growth companies but they don't have the kind of uh, hands to do all the things that they should be doing.

Speaker A: Yeah.

Speaker B: When it comes to whatever kind of area of go to market. Um, and of course both me and my, my co founder, my co founder, we both have pretty large um, networks I think both of us. My LinkedIn network is around 10,000 people or so. So we have pretty solid personal brands. So of course we're initially going to go, going to go probably around, around our own network. So that's Finland. But also because the type of company uh, that we are building it doesn't make sense to be only local. So we are going with, with the kind of international focus from, from day one, day one as well.

Speaker A: Very good. And in terms of, so you're looking at companies that have um, they're about to scale, um, and then you're kind of coming in and saying look, this is how you should be leveraging AI. Um, and we can help you do that.

Speaker B: Yeah, our um, model is a little bit different to a lot of, so there's a lot of SaaS around this topic now and you probably are aware there's a lot of like micro SaaS and then you have the big companies like HubSpot and Salesforces and others kind of providing this kind of full suite of SaaS. We are kind of leaning more towards the human in the loop aspect. And actually even though we are an AI native company, we believe that the human at the center of the operation is still the most important part of the puzz. So, so we, we use this um, um, growth strategist plus team of AI agents model. So what we, what we do is we embed a growth strategist into the client uh, company and then that person comes basically with the capabilities of leveraging a bunch of AI agents, leveraging a bunch of AI automations. To streamline the go to market operation of those customers.

Speaker A: Gotcha. And then so you would provide the, the growth strategist. No, he's like kind of say one of your employees, you'll put him in the company, they'll get an understanding in terms of what they're trying to achieve and then you'll build the agency orchestration kind of around that and then they'll manage it, optimize it, make sure that it's working to the best uh, uh. Capability.

Speaker B: Yeah, yeah, exactly, exactly. Because also a lot of the challenge why companies are not leveraging AI and not leveraging agent uh yet is just the lack of understanding of what's possible to do and also the technology kind of understanding. But there's also a lot of kind of rhetoric, tape legislative things, kind of privacy things that we like. I have a good friend who is working as a um, as um. Uh I think he's the CEO of their kind of agentic company and he said that we're actually a compliance company rather than a uh, marketing agency. So it's a lot of compliance stuff. I'm really feeling it right now when looking at our contracts for example and try to understand like what can we actually do and what needs to be approved by which kind of party etc.

Speaker A: Yeah, yeah, yeah, yeah. I think that's a big challenge. The further the size of the company is really kind of seeing that they'll share with you like kind of a data set. And then from what I've seen and I've talked to a couple of people in a similar space and they're saying to me that the big challenge is, is like the compliance piece. Um because a lot of companies are afraid of like kind of security threats and stuff like that. They know there's an efficiency there but they're maybe not as inclined to risk as maybe smaller operations are more agile and they can kind of see that uh the uplift is worth it. Whereas a bigger traditional company I guess um, they're more risk averse in that sense.

Speaker B: Yeah, yeah. And that's also in our thinking and my thinking right now is I think ah, we are the best suited for a partner like a mid sized company maybe even on the smaller size especially like growth companies like scale ups because they don't yet have that very rigid kind of enterprise company.

Speaker A: Yeah.

Speaker B: Kind of compliance and all of that because that's a whole different kind of uh, beast to tackle as an agency partner.

Speaker A: Yeah. When you look at then value propositions with companies like what are the typical things Because I think some companies it's like time, it's like okay, you've reps and they're looking for leads. Then you can build something relatively straightforward uh, from someone that's AI native, like build a clay agent and you can send them X amount of leads on slack a day or whatever. Now if you take the before and after that, that's like a huge time saving. If you're a rep and you have to go through Sales Navigator and you have to see okay is my ICP they engaged. Whereas if you automate that it's like they're the list, that's the leads and go after those. It's a huge time saving. But apart from that are there like other kind of trends that you see that like are kind of the main thing of like time savings or like any specific kind of tactics that companies kind of go like look, we didn't realize that AI can do this for us and now it's just a game changer.

Speaker B: Yeah, that's a great question. And I think the conversation around AI is focusing too much on time saving. I think time saving is like a clear uh, value, it's a clear value proposition, it's a clear value for the companies. But I think a lot of where we come in and what I've been able to build now in my previous role as well as now building for the new company is what can we do now with the help of AI agents, with the help of automation that wasn't possible before. So for example scaling on content uh, across multiple different platforms, kind of getting the visibility, getting the AI visibility kind of geo optimization uh, around there or basically adding something into our sales process that wasn't completely, wasn't kind of feasible to do like one to one personalization on the contact level things, things like that, that um, that just weren't even a consideration because it would just be completely kind of crazy to do manually. And then it comes a lot to do with how we leverage uh, existing kind uh, of first party data as well as public data. Public data is actually one of the biggest uh, like treasure troves when it comes to using AI agents. Because especially if a company like the target companies are even a little bit bigger, there's so much information around there of what the company is saying, what the employees are saying about things, what we can leverage as basically building battle cards for the uh, customers or building kind of ways in like conversations that you can leverage to get into those companies when it comes to sales for example. And then of course Automating stuff and uh, saving time is the kind of clear thing.

Speaker A: Yeah, yeah, yeah. And how do you find. Because the Nordics now in the last probably year or so there's a huge spotlight on success stories there from Sweden. Um, I also heard in Finland you have a really well known accelerator there. No, in Helsinki I think there's quite a well known one I've heard of a while ago that's got a lot of um, a lot of press about like a lot of success stories that have come out of there.

Speaker B: Yeah, yeah. Um, there's, there's a few in, in Helsinki as well.

Speaker A: Yeah. How is, how is it then the scene like in terms of the startup scene, do you feel that there it's vibrant? Is there a lot of kind of what would I say, like even um, incentives for whether it's like co working spaces but also in terms of how government kind of um, ventures, companies, agencies and stuff like that.

Speaker B: Yeah, I think I would personally love to see a lot more in Finland to be honest. I think there's a lot of discussion in Finland that we are kind of, especially when it comes to these kind of AI companies. AI native companies. I think Sweden and even Estonia is kind of pulling ahead. Sweden has you know, lovable. They have uh, Legora, these kind of companies. Huge, huge unicorn companies. We have our own in Finland and we have some very well known um, like marketing technology companies that have been around for a while.

Speaker A: Is Funnel Finnish Super Matrix?

Speaker B: Yeah, yeah, I think Funnel is Swedish.

Speaker A: Yeah, yeah, yeah, yeah. Supermetrics though, I know them. They're a LinkedIn partner.

Speaker B: Yeah, Supermetrics is one then you have Smartly for example.

Speaker A: Of course.

Speaker B: Yeah, there's a bunch of, bunch of others as well. So we do have uh, the kind of ecosystem um here. There's some big companies come out of Finland as well, like Vault for example. Huge company, huge company now. But I think there should be more. We do have like a good uh, startup ecosystem uh, in Finland but definitely, definitely could be more. I think government support is, is pretty good in Finland like in terms of like what that you can get what kind of services and funding etc. So, so that's, that's pretty good. I think it's more. But we are a small market like we are, we are such a, such

Speaker A: a population in Finland.

Speaker B: Oh my God, don't put me on the spot. It's like five and a half million

Speaker A: somewhere around the same as Ireland. Yeah, it's the same as Ireland. Population.

Speaker B: Yeah, yeah, exactly.

Speaker A: It was a little bit of rent. Same size.

Speaker B: Yeah. Exactly, exactly. Uh, now somebody's gonna make fun of me if I missed it completely. But it's around there.

Speaker A: Yeah, yeah, yeah. Ah, yeah, no, I had a feeling it was about the same size population, uh, wise.

Speaker B: Yeah.

Speaker A: So with, with um, your new venture then, how are you thinking yourselves in terms of go to market, like client acquisition, things like that. What's like in. And I know it's like kind of when you're starting something, you're going to try everything and see what works. But what's your initial intuition telling you? That would be like kind of where you're going best return?

Speaker B: Initially, uh, two things, uh, that's actually really clear. I think for me it's building public. I'm a big proponent of building public. I love building public companies. There's a great one in Sweden I can shout out definitely, uh, called Fibler. I don't know if.

Speaker A: Yeah, yeah, yeah, yeah. I know them. I had Adam on the show a while ago. Yeah.

Speaker B: Oh really?

Speaker A: Okay.

Speaker B: Yeah, yeah, it's, it's, it's funny because there's two Adams actually.

Speaker A: Yeah, that's right.

Speaker B: Yeah, yeah, Adam and Adam.

Speaker A: But yeah, ah, yeah, yeah, no, they do it really well.

Speaker B: Yeah, exactly. And I think that's a great kind of benchmark for me, like how to build like a great go to market system and build it in these kind of steps, like focusing on a couple of things that make sense and work really well. And then once you have those dialed in you kind of add the next layer like they're going into CTV advertisement to YouTube. But yeah, building public is definitely number one. Uh, and then personal brand is the second one. I think we have two very strong personal brands, me and my co founder, um, leveraging that because that was also one big part of uh, our GTM at Guru was around personal brand and thought leadership, total leadership ads, uh, around there. So I think that's going to be one. But I think then, then I have this challenge with the go to market is because um, I think there are things that I'm not even thinking about. I just had this idea I think yesterday is um, because a lot of companies are putting out these, you know, freebies like Claude skills, Claude plugins, stuff like that and they're kind of building in marketing into those skills and into those kind of repositories and stuff where if you build something that is really nice to use, something that I use personally in our kind of go to market and kind of make a public version of that, um, building uh, in a kind of way that hey, if you need extra assistance like here's contact information or stuff like that, I have an example because I have built a really good like a journalist agent for myself which I use as the basis of my personal kind of uh, posting on LinkedIn. So basically the journalist, it interviews me every week. So I have this session where I have my running pad here in my office and I talk on my running pad. I put on Whisper Flow, which is an amazing.

Speaker A: Yeah, yeah, I know it too. Yeah, yeah.

Speaker B: So I put on Whisper Flow and then I start brain dumping everything. It asks me follow up question. It does this very cool thing where it pushes uh, back on my thinking. So I have built in this, I call it the Steel man because you know, straw man. Steel man. So Steel manning an argument. It's, it's making kind of the opposition's point. So I have the Steel Manning built into it. So I get really kind of robust uh, posts out of it. And then it has a copywriting agent, it has the publishing agent. So I use everything like that. And I've been talking about that publicly on LinkedIn. So many people were so interested that I just made it kind of publicly available for people to download basically like a template and a skill kind of repository how to use it. And then afterwards I was like, why didn't I uh, build in like marketing into that thing itself? Because then it's still kind though I'm going to add it now after this discussion I think. But these kind of things I'm thinking like as an, as an addition.

Speaker A: How does it work then in terms of. So you built the skill on cloud code? No. Yeah, um, you have it connected, you use Whisper Flow to give. You basically dump down the things. Then do you have it connected with Whisper Flow to cloud code or is it more you just jump it onto like a kind of a file then afterwards based on what the output is on whisperflow?

Speaker B: Yeah, I basically use Whisper Flow to transcribe everything directly into the cloud code agent. Um, and then I have the connectors to um, I use a tool called Xerneo which is like an API for all social media platforms. So it can publish automatically through there. So basically I have like a database and then it has this kind of learning loop where it pulls all the analytics from all of the posts and then updates the copywriting agent, ah, into like how to talk about things and what angles to take and stuff like that.

Speaker A: And will that give you like a straight API onto LinkedIn? So it'll basically just post the draft there or it posts it directly as well, if you want.

Speaker B: Yeah, it posts it directly. This was recommended, uh, from uh, an acquaintance of mine. They run this um, like LinkedIn, ghostwriting, uh, platform and service. And he said he recommended this tool, um, to me and it's LinkedIn and it's all other social media platforms. And the really cool part, what's the name? It's called Xerneo with a Z.

Speaker A: Z, E, O, N, U. Okay.

Speaker B: Yeah. Z E, R I O N. Zern

Speaker A: I O, N. Ah. Uh, I think I know. Are they based out of Barcelona?

Speaker B: No, I think so. I think it's a European company.

Speaker A: Yeah, yeah, yeah, yeah. I think, I think I know them. I think I know who it is. I must check afterwards if it's the one I'm thinking of. That's handling.

Speaker B: It's very cool. It's very cool, especially the analytics part because a lot of the tools don't provide you API access to the analytics and this one does.

Speaker A: So it will give you the LinkedIn analytics on all your posts and then you can feed that back into a learning loop. That's very useful.

Speaker B: Yeah, it's very cool. It's very cool. I have it run every time I do this kind of weekly content production and content update. I have another kind of sub agent run that kind of audit, uh, while I'm doing it, then kind of pull the information backwards.

Speaker A: Yeah, that's really handy. How do you find then, um, with like, do you use uh, uh, it's for want of a better word. But I suppose this is what like the general term in AI is called a second brain. Um, I don't know if you've like kind of um, built. How do you find. If you were to talk about like before you had something built like that versus like now, maybe you could just give us a little bit of an insight in terms of how it's improved your day to day.

Speaker B: Yeah, absolutely. I think everything what you do with AI, especially with agentic AI, so you have autonomous agents running tasks. The data foundation is everything. Like that's where things kind of succeed or fail. And I think that sort of second brain or context layer or whatever you want to call it, I think it's very, very important. Uh, because then I'm able to. So if I have an idea, just yesterday I had an idea to kind of create this uh, backlinking agent for SEO, SEO purposes. Uh, I can just give it the context. Like, hey, here's everything that's already existing in the tech stack. Here's all the information about our company, our target, like what types of things, so it can pull all of that information and then it has basically the stuff that would be like a uh, kilometer long prompt for me to begin with. And of course then things change, especially in a new company, things change very quickly. So actually managing that sort of business context, like, hey, we don't actually use this tool for this anymore. We use this other thing now. And of course in our kind of model we built that kind of second brain for our customers as well. So that's kind of the core of our model because everything runs on top of that, that information.

Speaker A: Yeah, yeah, yeah. And I think once you kind of get that in and you see the before and after is huge because you can kind of see that it's self learning then as well and it improves based on the input that you get.

Speaker B: Yeah, exactly. But that's, I think that's where a lot of people using those kind of second brains kind of can fail is a lot of that stuff doesn't happen automatically. So you actually have to build in the infrastructure for the system to update its memory and update its information. So I for example have this little skill that's basically like a recap skill. So I can go recap on um, Claude and it will look at the whole discussion in that session and see what it needs to update into the second brain.

Speaker A: That's really useful actually. Yeah. Uh, and how did you create the skill? Just did it on Claude. No, like a recap. You just put in the prompt and it created it, right?

Speaker B: Yeah, exactly, exactly. I have a separate skill for building skills. So it has the information like how to build a skill in a way that I want it to be built. I have like um, in my second brain I have this uh, like a design philosophy which is basically using Anthropic's design philosophy. So Anthropic has this kind of publicly available information about how they designed things and I use the same logic. So there's for example when it comes to AI agents, you want to have it's, it's funny, it's this, it's uh, we call it uh, fat skills, thin harness. So you have a very robust skill with a lot of information and a very thin harness. So not a lot of um, APIs or technology to ah, achieve the thing. So that works quite nicely. And I try to follow the same kind of build logic across all of the different things because if things are built in a different way, they might not kind of work uh, together really nicely.

Speaker A: How do you find it then with like using say LLMs and like, like I think in Europe it's probably say, fair to say that everyone's goto now is probably Claude code. But yeah, as a result of that there's, there's um, there's a cost that's always involved there in terms of token costs and they're only going to increase as the company begins to move towards profitability. But do you see like kind of a way of like even kind of like having your architecture, um, whereby you know you're, and I know a lot of people are moving towards this kind of like efficiency with tokens, but not just being reliant on one LLM to try and kind of structure it in a way by where the menial tasks are using something kind of cheaper, whereas if it's long learning, it's using some maybe more advanced LLM. How do you think about that when you approach it?

Speaker B: Yeah, exactly. Exactly like that. And I have this again in the actual kind of design philosophy. Whenever I think of a task, uh, or we think of a thing that needs to be either automated or added, uh, kind of generative element, there's these kind of layers. Something doesn't have to be an agent. I think that's also one thing where a lot of people kind of um, go wrong is that you, if the task is um, if the task is not generative, it's deterministic. So one plus one equals two. Or put this file into here. You don't actually have to use an LLM and nor you shouldn't use an LLM LLM. So then we go into just basic automation stuff. So for example, I have used uh, N8N for a really long time now, uh, kind of to do a lot of those tasks. So then you choose. Okay, this is actually just an automation. It doesn't have to be an LLM. And then something that needs to be, it's non deterministic or needs to be generative. Uh, then you should put an LLM into it and then you. Okay, is it just a single API call or does it require an agent or does it require like agent orchestration? So multiple different agents doing a thing and then you go into choosing which model works the best. Do you use Fable for everything, which is extremely expensive or do you use something else? But it's super interesting uh, because I think especially big companies M. There was this crazy story, I think from the US there was a consulting agent, consulting agencies that got a $500 million bill from Anthropic.

Speaker A: Wow.

Speaker B: So it's probably one of the big four or something. You know it has to be.

Speaker A: Yeah, for sure. But like. Yeah, it's incredible as well because I think there's talk now that like in larger organizations, engineering companies, their KPIs are based on how much tokens they're using. So they really want to make sure that they're. Whereas that's not necessarily efficient as well because if you have it hallucinating it's going off the guardrail, it doesn't mean that the output is going to be better. It might just be like kind of off tangent a little bit.

Speaker B: Yeah, yeah, absolutely. And I think, I think every company is going to have to start thinking about where do we actually implement AI and how much do we actually like have it do. What, what, what makes sense. I think the real kind of cost struggles is really with the, with um, software development companies and kind of coding. I think currently I don't think in, in marketing and sales use cases we're going to have a problem with the cost. I think the benefits are so great compared to the actual costs because like for example I have a bunch of agents that uh, run automated reporting, run automated kind of prospecting stuff and those and we're talking like less than like €1 per run.

Speaker A: Yeah.

Speaker B: So it's, it's just me spending two hours to build it is more cost than the whole thing will cost in a year or two. So.

Speaker A: Yeah, yeah, you just mentioned something really interesting there for like kind of um, say automation for like even say cold um outreach or like what do you think in your opinion? Because now a lot of things are democratized but also it's probably maybe more challenging to reach people in decision making processes because there's so much, much information out there. There's also then so much automation with like sequences and so on from what you see. What do you see that actually cuts through the noise now versus probably didn't before. Um, or that doesn't work now and did before.

Speaker B: Yeah, I think that's an interesting question. I take maybe a little bit like a contrarian approach to kind of outreach and using AI. I think like how I used AI, uh, agents and AI processes at Guru for Sales was more around automating stuff and kind of freeing up time so that I can spend more time in like high quality one on one situations. So actual physical events and meeting person for meeting people for lunch and stuff like that. So I think, and I do see that as a trend in marketing overall. A lot of marketing agencies are hosting, you know, breakfast events and dinners and so it's nothing special, but I think um, just trying to automate outreach, trying to automate your kind of sales outreach, especially for an agency, I really don't see that working really well. And I've tried all kinds of stuff myself but I don't really see that working all that well. I think where AI becomes really good is picking uh, up on signals. Um, and something that I've used it for quite a lot is understanding like basically lead scoring. I don't really like the term lead scoring. I don't think lead scoring as the kind of old kind of quote HubSpot way of doing things kind of makes sense. But I think just understanding who are the most active towards us, like who should we reach out to, both us in companies and individuals. And there you come into like Fibler. For example in my Google Guru stack we used Fibler for that to push in account information which companies are the most active in our thought leader ads to understand who could we reach out to what make the most sense. I don't think it's increasing the number of activities is not the goal. I think it's increasing the quality of the activities that we decide to do is the goal.

Speaker A: Yeah, which makes a lot of sense because even now uh, you're probably aware of pavilion, um, the SaaS kind of, of, I suppose they're like a SaaS community. They brought out a report there a couple of months ago and they were just talking about like how you know, it's really challenging out there but the companies that have like much higher activity, it doesn't necessarily mean that that contributes to like more one close deals. At the end of the day, the ones that are actually uh, routing the deals to specialists to really, that are suited to the vertical to the particular sector, those deals are much more likely to close. So it kind of gave this thing of like how you build a revenue orchestration around that. So like with AI, uh, like what you mentioned, when something comes in it goes to this particular contact and it's not just like whoever's available kind of thing. So I think that seems to be the place of like really knowing who the customer is, but then also making sure that when that opportunity comes in that the right person is talking to them. And that's difficult to do. I think the bigger scale that you get, uh, as well.

Speaker B: Yeah, but I think that's a perfect example of that sort of like organizational level context and managing the organizational level data. So as an example in that sort of case, what uh, I would kind of build would Be understanding who has worked with what type of cases. So having like um, a database, maybe a vector database or something like, okay, this specialist has worked with you know, aviation and railroads and stuff. And then when the lead comes in and there's you know, analysis of the company and analysis of the message done automatically by an LLM. Okay, hey, this person has actually worked with a lot of these cases. Let's route it to this exact person. And that's the type of thing that uh, doing manually would have been just crazy to do before. Yeah. Now it's just almost arbitrary. Not quite, but you know.

Speaker A: Yeah, for sure. The tool that you mentioned that you built yourself like the reporter agent, how does it work in terms of um, how you interview yourself? Like you basically put the dump into, into whisper flow. You put in all your ideas. Um, does it then like interview you verbally or what way does that work? Or is it more like it just asks you text questions and you kind of put it in and then you, you manage it then in terms of the output.

Speaker B: Yeah, text questions. I've had the idea to plug in like an 11 Labs API for, for Voice, but I, I just hadn't got around to. And it's a little gimmicky. On top of the thing, there's, there's a couple of, couple of tools. There's actually two people that I know in Finland have built this sort of thing that is like a discussion, uh, like you have an actual voice discussion. But it's so, how should I put it? So it works in a way that, okay, I, I do my brain dumping at the start.

Speaker A: Yeah.

Speaker B: Then it looks at my personal brand, uh, like a markdown file of my personal brand, my goals, who I talk to. And then it looks at the kind of uh, results of previous kind of posts. Then the analytics data comes in like what has actually worked before. And then from that very long thing it picks up, okay, hey, this part is actually similar to the thing that has worked before. And then it asks me follow up questions to get more kind of meat around the bones on um, topics. And then it splits it up into posts for the, basically the whole week. So hey, let's talk about Monday. Let's talk about this thing. Then we work on that post. Then let's talk about Tuesday. We work on this thing. And then I have another agent which does social repurposing for threads and X. So it takes like snippets out of the big kind of thing. So this is, this is taking from, I think somebody calls it the big Rock. Yeah. So you build like a um, yeah,

Speaker A: uh, and smaller rocks then.

Speaker B: And yeah, exactly. So, so that has been my kind of go to go to.

Speaker A: I found it overall then with having that like of. Because before if we're having this conversation probably about two years ago, it's a thing of like you go on LinkedIn and you, you write and, or you write on a piece of paper or whatever it may be. Um, I think the challenge, I don't know if you find this in LinkedIn which many face now is that it's democratized a lot of things so you get a lot of like kind of AI slop in a way of like hey, these uh, things for Claude, um, comment below and I'll share it with you. And they get a huge amount of um, still even they get a huge amount of um, engagement. I think in the long term it kind of ruins the personal brand to some extent but still they still seem to get a lot of engagement. But since transitioning over to where you have a solid system in place, you have your own points of view that are unique to you. Have you seen like, like even with the updates on LinkedIn, ah, like once you have that kind of structure in place, are you getting the engagement or better than what you got before or do you think it's still like kind of, it's still more challenging on LinkedIn because of those external kind of things of how the platform has changed and stuff.

Speaker B: Yeah, I think that's very interesting. I, I, I, I see that you are very active on LinkedIn as well, picking up on those kind of, those kind of trends. Trends as well. Um, uh, I was very, actually surprised because uh, ultimately I built, I have, I'm very strapped for time right now, everything so I really, my idea was like okay, I want to spend at, at maximum two hours a week for a week, weeks worth of content across all social media and my website. Um, so, so I, I wanted, that was the kind of initial idea for the build. But actually I was very surprised that uh, the things that the uh, posts that are coming from the journalist and the copywriter agent are actually performing better on average than what I would do just like manually writing stuff.

Speaker A: Wow. Right, right.

Speaker B: And I think it's, it's, it was very interesting. Very interesting. So, so they have done very well. I have a very kind of, maybe a distinct way of writing but I think one core thing is I had the audience before the explosion of content, uh, like the content on the web. So, so actually the personal brand from before because I'VE been doing this very actively for maybe four or five years now, like the LinkedIn content. So I think that is actually the most, most kind of important part, why it's performing still, like very well. And I think those kind of, um, transactional plays. I'm not the fan of, you know, the comment Claude for xyn, X, Y and Z. Like, I think if it's of actual value, it's great. But every single time I comment Claude in a post, I get something that is completely like, it's just automatically generated by Claude, so there's no actual human thinking involved at all. So that's, that's also like. And I do, I do agree that I think it, it makes your personal brand a transactional brand. And I don't think.

Speaker A: Yeah, no, no, it has a shelf life that's not very long, I think at the long run. And like, no matter what you do in future, I think if you have that, it's kind of, It's a bit spammy. Yeah, yeah, yeah. And what else? I'm, I'm all these things are coming into my head to ask it because the conversation is really interesting. But, uh, what about tools that you see like clay? Um, some of these, really, it's, it's interesting because like, there are other tools that are like clay out there, but I think clay have done it really well with the creator space and also how their founders. You mentioned about Adam, and Adam has done it really well in Europe. I think the clay team, obviously they've, they've raised a ton of money so they can do it easier as well because they have a lot of investment behind them. But they've seemed to really, um, capture this thing and even created their own category in a way of like kind of GTM engineering that they've coined. But if I, I don't know if you've used the tool. I've tested it a few times and yeah, it's good. I think there is, there's a lot of value there. Um, but I also think on the flip side of that, it's kind of, if you're just using it to be really sophisticated in terms of automation, I think there's some, some, um, uh, there is some limitations there. But I'd love to see from your perspective what you think about clay and from your experience, do you think there's a risk there that like, people that lean in too much into that automation and kind of go with the whole getting behind the tech stack of trying to automate everything and not really having like, what we talked about at the beginning of having the human in the loop and thinking like that you can just scale and scale and scale but with the further you get away from the customer then the risk is that you're not really have a finger on the pulse of actually what's resonating versus what's just pure noise.

Speaker B: I think that's so true and I think the whole concept of go to market engineering, I think I posted about this uh, some time ago on LinkedIn, is that it's kind of the tech bros discovering marketing. Yeah, it's honestly it's just marketing. It doesn't have to really have a different kind of word or marketing and sales but you know, that sort of thing. I think it's completely true that if you think that it is just a process that you can engineer, I think that leaves out one really, really important part which is the human at the center, both and the creative operator and the creativity and all of that. So I think somewhere in between is the sweet spot, spot. And I'm trying to find that myself like for our own go to market and trying to understand like how much I think of AI as this gauge all the time is for any process or any task you gauge like, okay, does it make sense to use 30% of AI to do 30% of that or 50% or 60 or 70 or 90? But I've yet to find a thing where 100% really makes sense.

Speaker A: Yeah, yeah.

Speaker B: And I agree. And then, yeah and then it's maybe just automation. If it's 100% then it's automation and then it's not AI at all.

Speaker A: Yeah, yeah. And I think also if you scratch beneath the surface like some of these companies that are, you know, you've probably seen some of the um, posts like I have this company and it runs by itself and we've generated X amount of millions. Um, uh, you kind of, you, you can kind of see when you scratch beneath the surface it's actually kind of like a bit of a Ponzi scheme in some senses. Like there is definitely there's, there's an element of that to, to the sensationalization and when you see it it's not really, actually realistic. And I think that uh, there's definitely an element of that that kind of calls things into, into, into question as well.

Speaker B: Yeah, absolutely. I think it's. And, and, and I think, I think that's also the conversation has been so much around the time savings and automating things, but not enough about the actual output. I think I See that, Because I'm like, I've worked with SEO for 15 years, for example. I see it a lot in the SEO space where, uh, a lot of people are coming. It's like, hey, I've automated my whole SEO to Claude. And then if you, you exactly scratch a little bit beneath the surface, it's like, actually your results are the same or worse or really kind of worse. So it doesn't make sense. And I think that's also to the idea of our new agency and our new idea, the human in the loop, uh, in that is that, um, where we get the most benefit out of AI is having a subject matter expert using AI. AI?

Speaker A: Yeah.

Speaker B: You, you don't become a subject matter expert in SEO. You don't become a subject matter expert in, in sales automation or marketing automation. If you have AI, you become mediocre at best. Like, I don't become a, I don't become a software developer, uh, when I have access to Claude writing code for me.

Speaker A: No, it's true. Yeah, that's for sure. It's really kind of. That's a really important point to point out because I think some people kind of think that you can automate it are you. And AI can do everything. But that really leaves where the probably the competitive advantage is based on how you can think about things differently, critical thinking, so on. Um, by not doing that, you're leaving what the real possibilities. Like it's a good sparring partner and stuff, but it's not something that you should just kind of outsource all your capabilities too.

Speaker B: Yeah, absolutely.

Speaker A: Uh, it was great having you on the show. I really enjoyed our conversation. If anybody wants to find you the new company that you've launched, um, where are you most active? Where can they find you? LinkedIn.

Speaker B: Uh, for sure. Yeah.

Speaker A: Okay.

Speaker B: Yeah, Everything will be, Everything will be there once. Once we have that new company going live, uh, in August.

Speaker A: Perfect. We will put the. This will probably come out after August anyway, so we'll put it live in the, in the show now notes if. Do you have a name for the company actually, so we can, we can put it in the show notes as well.

Speaker B: It's a, It's a secret.

Speaker A: Okay. All right, good. Suspense is a good thing. Uh, perfect. Well, listen, thanks for joining us and, um, it was a pleasure speaking with you.

Speaker B: Yeah, thank you so much.