DN #22: AI Workers, Lean Teams of 50 Agents & Why You Can't Delegate Strategy (w/ Davida Ginter)

Niklas:Hi and a huge welcome to you my lovely listener. So glad you're here. Today you're joining me for a chat with Davida. Davida is a 3x founder and now leading Elo. Davida, so nice to have you.

Davida Ginter:Great being here, thanks for inviting Minikas.

Niklas:When you, maybe a few words regarding yourself in the beginning for the people to introduce themselves, how did you get started as a founder?

Davida Ginter:I started as a founder, actually it was almost 15 years ago, 14, 15, something like that. ⁓ Before that I was a journalist and although I liked it a lot, I kept, you know, envisioning myself actually doing the things and not just writing about, which kind of led me to a journey. ⁓ Back then I had a services company around burnout prevention and then it turned into a startup, my previous startup and now a bigger startup, which is Ilu. So yeah, the journey about becoming a founder was really started as the idea of how can I, you know, just build something myself, make an impact as big as possible.

Niklas:When you started as a journalist, what did you bring over? What do you think were the really valuable traits that you brought over to being a founder later?

Davida Ginter:That's a good question. You know, I think many... And rightfully so, think that journalism is about good writing skills, which oftentimes it is, but the real skill of journalists is the ability to listen, to find the essence, what's important, know, listen to people, extract the essence and translate it into a very compelling message. So I basically use everything I just said as a founder, know, listening to the market, what people need, what users want, how can I build something impactful and then How can I lend it, creating the messages, reaching out to people. Distribution is as important as building the product itself, obviously.

Niklas:Yeah, I think it's even becoming more important now because it got at least for software products, it got so easy to build them now. So ⁓ I think a lot of people struggle with that, which leads me to a second point. I think burnout is a very common problem with founders who work a lot if they get stuck. You wrote a book on burnout prevention. I think that's really interesting. Why did you do that?

Davida Ginter:Yeah. Yes. It all started almost, you know, coincidentally, I had a conversation. It was 2016, 2017, end of 2016. I had a conversation with a colleague. had my agency, we're doing training and she led her own thing. So it was all around. leading behavior change and stuff like that. And we were talking about some other colleagues of ours, founders and change makers, and we felt everyone is feeling so stressed. about solving problems at a large scale. And ⁓ we said something in a conversation to the extent of you don't have to suffer in order to help alleviate suffer, right? You want to make an impact, but you don't have to kind of burden, ⁓ you don't have to feel so stressed out by carrying this mission. And with that idea in mind, thought, well, maybe I could write an article about that. You you can take the person out of the newspaper, but you can't really take out the idea of you want to keep writing about things. So I decided to write an article about burnout and I reached out to people to tell me their stories. And before I knew it, stories about funders experiencing burnout poured at me from everywhere. So the article became a book and the book became a company. which we led training around the prevention. Really it was just looking around and suddenly noticing that people are so ⁓ either stressed with the mission, but it's more than that. For founders, for change makers, this is about when you identify with what you do so much and you can't separate the profession from your own self. you can easily get burnt out. And I talked to people who experienced failures in the business, which is very natural, right? But at some point they said, I am the failure, which is a very, very dangerous place to be because then you get burnt out in this situation.

Niklas:What is the number one myth about burnout that you wish more founders knew? What can they do when they feel they are getting in this situation?

Davida Ginter:There are several things, I people think that burnout happens because of long working hours, you know, and but this is not true. Actually, I know people who work 12 hours a day, feel ⁓ very satisfied, they're not even close to burnout as opposed to people who work just part-time and feel burnt out. So actually the first or you know I don't know if it's number one but at the top would be more about if you're pursuing a mission you're not truly believing you will get burnt out. If you ⁓ put a lot of time and you don't have the right resources. So it's not about the time. It's about having the right resources. You will get burned out. and, and by the way, also, if you don't balance the professional needs with the personal needs, you will get burned out. And it's not about work life balance. It's more about what's currently missing in your routine. So if we're looking at those factors, the advice for everyone, but founders in particular is to honestly look at our own calendar and time allocation. and understand what are we missing there in the balance. Should we socialize more because we spending time alone in front of the screens? Should we find more physical activity? Should we have some more quiet time for reflection? could be any of those. doesn't matter, right? As long as it's balanced. But it's also about can you go and find the resources that you need to accomplish your work? If it's not balanced and you are pursuing something without having proper resources in place, you are burnt out because you're trying to pour from an empty cup. It's just too much for one person to do.

Niklas:Yeah, I was wondering about that because when I came out of after my PhD, I went into consulting. Consulting is known for long work hours and I was never anywhere near burned out. worked a lot. that might be different for different people, but for me, definitely there was this, I cannot find the correlation between that and burnout. So it's probably more about

Davida Ginter:Yeah.

Niklas:protecting one's energy or something like that. Do you have an idea how people can do that well?

Davida Ginter:Yeah, it's definitely about protecting the energy. The best idea to do it well is to, well, there are different, ⁓ one exercise I liked doing with founders I met ⁓ was just open the calendar for and look at the past four weeks, eight weeks, even the last quarter. And first mark all the professional activities that you do for your business, for others and so on. And then mark everything you do that doesn't give you energy and you don't even understand the purpose and finally mark the things that you do for yourself and actually give you energy and then just look at those three columns and See if they are balanced in most cases. It's it's not balanced not even close to something that you can ⁓ Rejuvenate and get energized by and look at that and suddenly realize I spend 10 hours a week doing things I don't even believe in or don't contribute to anything and just you know degrade my energy ⁓ or ⁓ resourcefulness then you realize I need to change something in my routine. So it's not about working less hours, it's about what can I incorporate in my routine that will actually give me this energy to keep pursuing my mission. So just looking at your calendar for the past few weeks will suddenly reveal amazing things.

Niklas:That's maybe an interesting tip. then currently, to do the switch to your latest startup, you're now building a digital business. I'll say again. ⁓ What is ELO about, if you sum it up in three sentences?

Davida Ginter:Yeah, Ilu is a platform that enables business owners and particularly non-technical business owners to operate their business in a much more efficient way by using AI workers. ⁓ could be agents, could be complete workflows, but the idea is that you let Elu, the platform, understand your business. You share some context, you let Elu connect to your tools and Elu will build for you. a gigantic workflows that can take on operational work, go to market work, any repetitive work to enable business owners to focus on human relationships and do what they really excel at in their business. So if you think of it could be any professional agency, right? It could be real estate agencies, recruiting agencies, could be home repairs. have ⁓ architects and they want to sketch their architecture drawings and they want to sell houses and they want to recruit people and they want to, you know, do home repairs. They don't want to do all day long the 80 % of the operational work and go to market work. And they would like to delegate this to AI, but they don't know how because they're not technical. And so our platform literally builds these AI workers for them. to take on the heavy lifting of the operational work and some of the marketing work and some of the customer support work that takes so much time of the day from those business owners.

Niklas:And how can I imagine that if I sign up for the first time, how do I work with it?

Davida Ginter:Yeah, so some people sign up. ⁓ We launched just a few months ago, we already passed 300 businesses. So some people sign up and immediately jump in and start describing, I want this type of agent or that type of agents. And they can immediately start operating these small agents for the benefit. While others, want kind of the fullest version, which we are currently in closed beta, launching very, very soon to more publicly. And within the bigger version, this would be about just sitting there for an onboarding process, sharing the business context, you know, just telling the platform what is that you do, what's your business about. And from that moment on, will proactively start suggesting you what agents you can actually create there, or it will create this for you, but you want to approve first the agents. ⁓ so understand, for example, ⁓ you have an agency that ⁓ I don't know, help technicians do their work, right? And so, Ilu will come up and say, maybe you want an agent that will handle your customer support. Should we build this for you? And you said, yes. And Ilu will actually build this for you as a platform and so on and so on.

Niklas:Yeah, I think that's interesting. And for people to understand, so agents then will do some kind of autonomous task. So ⁓ I would describe the agent. It would start. What do you feel over the last months? Have the models still gotten better? Are your agents getting better with the technology in the background? Or is it more like stale at this point?

Davida Ginter:It's a very good point. I keep telling my team that we are building not for today's technology, but for tomorrow's technology. So the way we structure the platform is that there's a very... ⁓ Simple interface. I remind that our users are very not technical, salespeople, operation people, people. So the interface is very, very simple, but under the hood, the engine is all about the agentic workflow that we have kind of coded in. And so the more technology advanced, obviously the agents get much, much better, ⁓ better outcomes, better guardrails. are better able to self correct themselves. So we are actually counting. on all technology to become so advanced and so much better that our platform would only benefit from any improvements out there in the market with the infrastructure.

Niklas:When you look at users today and how people use AI, what are typical mistakes that you see? And where does ELO step in and change that?

Davida Ginter:Yeah. So sometimes people expect magics without actually describing ⁓ in more detail. ⁓ I'll take this to the analogy if you're now onboarding a new employee, a person, know, human. So if you're onboarding a new employee, even if it's the most talented person, you know, very experienced, if you don't explain to them what you need and give them some context, ⁓ you can't just, you know, send them to do the work and expect, you know, they will create something wonderful on day one. You need to give them context, you need to give them some instructions. You actually want to receive their first outcome and give them some feedback, some review. And the same goes for agents. The more involved you are, the more you can expect something accurate and not just magic. AI is not magic. It's artificial intelligence. And that means that it should be intelligent by understanding what is that we need. The way ILLU approached it differently is that we noticed that many non-technical users, they don't know how to prompt well, how to put the guardrails rail. And that's fair enough. That's not their job. So we abstracted at this entire technical complexity to enable our customers, our users to describe their business goals instead of prompting. So instead of going in and said, I need an agent that runs twice a day and do XYZ. They can go, here's my business context. Here's what I do every day in my business. And here are my business goals for the coming quarter. What can you do for me? And from that moment on, it will create the prompt. It will instruct the agents. and the business owner's responsibility is to give it feedback. Is this going in the right direction or not? So this is how we of shifted from prompt team and specification to business language, KPIs, goals, et cetera.

Niklas:I've seen a lot of these, I would say, ⁓ local builders where you build workflows. I think your approach is a lot about describing the agent or describing the task. How does this workflow work for you, and why did you decide for this work to get the other approach?

Davida Ginter:Mainly because our audience or we decided actually to give service to over 90 % of the market who don't want to be absorbed by technical complexity. So this is where we start. This was our starting point. You have the builders, they like Claude, Claude rightfully so exactly. But the other part of the market, is huge share of the markets, they will not open a terminal now, right? They will not go and fetch an API. They don't want to go into thinking about algorithms and two details of a workflow. So for us, when we started thinking about this audience and what do they need, know, business owners, not technical teams. The approach was about how can you continue focusing on your business? Even when you work with Ilu, how can you describe the business context and your goals rather than focusing on building an agent? The purpose for them is not to build an agent. The purpose for them is to hit their KPIs.

Niklas:Yeah, that's true. And if you think about as easy as possible, you probably thought about it. So you came up with, just need a text box where I describe it and everything else happens in the background. Or is there some step in between?

Davida Ginter:Well, steps could be, example, so mainly it's text box or conversation and some clicks to connect the tools and the rest. Yes. It's done autonomously by Ilu. Still, there's a lot of room for business owners to be involved in the workflow. So they can say, I want it fully autonomous. Right. But others will say, no, I want those review points. I want to approve every piece of content that the agent will create for me. Great. I think this is a decision so they can do this. I want to give the agent feedback so that next time it's even more accurate. Great! mean ⁓ they get to choose how much they want to be involved but they are involved not in rebuilding the agent, they are involved in giving it feedback just as you would work with this new employee that I mentioned before or even not the newest employee. You want to give feedback to every one of your employees not just the newest one.

Niklas:And who is your typical customer? It's probably, is it large enterprises you're not targeting smaller ones, I think, currently?

Davida Ginter:No, definitely smaller ones, small to mid market. A lot of professional agencies, a lot of services agencies. could be small business owners, could be much bigger ones with 40 people on the staff and doing, you know, a few dozens of millions in annual revenue. But it's all about people who are knowledge workers and giving services to the markets, not large enterprises.

Niklas:I think what I've been discussing with a few people, what I feel is juniors in a lot of especially service businesses can be replaced by something like ELO. When I go back into my early consulting days, I built a lot of PowerPoints, Excel models. ⁓ If you look at the tooling and also what you're building, a lot of that is going away. I always wonder what will the future look like from your perspective in this point and how will it?

Davida Ginter:Yeah.

Niklas:change the also maybe hiring landscape.

Davida Ginter:Yeah, I think about this a lot about this very very specific question about this very point. So are we going to a future where Everyone are solopreneurs. I don't think so, right? But you know, some people will claim anyone can now be build a billion dollar business with some AI I don't think it's true and I don't think it's going there. I do think that we are going to see three major shifts. One is much, much smaller teams, lean teams even like. ⁓ three to ten people building great companies with the staff of 50 agents, but still three to ten people who actually manage the entire strategic direction and the human relationship. You can't delegate strategy to agents, you can't delegate human relationship to agents. Agents can do pretty much everything else, but ⁓ when we're talking, by the way, about knowledge workers, I'm not talking about the physical or hardware. even though that could also be delegated to machines, right? But we're going to see that the first shift is much, much smaller teams, lean companies doing very successfully in terms of, you know, impact and revenue. ⁓ Another shift that we're going to see is a replacement of the old model, like the ore chart. So we're going to see a lot of consultant that are actually like an AI for let's say AI accountant and AI legal and AI ⁓ HR recruiting and you don't remove the person from the company but you see that they are leaning heavily on AI to scale the business and how actually they operate and give service. And probably the last one or kind of the last shift is that we are going to see a lot of the work before. Now we see a lot of automation of current processes, right? I think that very, very soon people will say, why are we automating something that was not efficient to begin with? And we will start seeing people rethinking the actual processes, the actual workflows. So instead of saying, today I'm doing cold emails, so let's automate that. It will become, can I get the market or can I do run operations in a different way to begin with? And A, I will carry that. So I think we're going to actually rethink. how work is being done instead of now we are all jumping into automation, but maybe we are automating the wrong thing.

Niklas:So I don't know what the result will be, but I also believe that we will probably see productivity increases before we see these large shifts towards smaller companies. I think it really depends on the type of company, obviously. If you are, your only job is to produce software. As a company, as a service provider, you will need a lot less people and probably the quotes that you will get on the software you build will be a lot lower. I think that's clear and that will happen already. But if you look at large enterprises, there have always been trade-offs. Even in the production, you mentioned it in hardware. If you look at a production environment, lot of things can be automated and could have been automated for years. There's just a trade-off in costs because if you have a robot, you need maintenance as well and it's not flexible. So if you do the trade-off calculations, oftentimes it's just cheaper to have a human worker. That's the truth. I think we find something in that ballpark also in knowledge work. There is, that's at least what I guess, ⁓ a gap in between where you have roads that will remain if it's really repetitive like today entering. data into an SAP, for example, or some other ERP system, whatever you use. These kind of things, I'm not sure if they will remain. think they will disappear.

Davida Ginter:Yeah. Yeah, they will disappear and I think we see more and such just, you know, this week, so many layoffs or last week, so many layoffs have announced from corporates, right? And, you know, Meta is firing, zoom in for now weeks and, know, into it and... Part of it is because there was this assumption, okay, so if we just add some AI features, right, throw AI features at this legacy system, we will solve the problem. No, I mean, it doesn't work this way. So if now you're not much more efficient and AI native, there's not much you can do with this, all this fact you've built in your company for so many years. So I'm thinking about, you know, this shift you're describing, it's going to be... not just about hiring and people, it's an entire economy shift of how we do work, know, what's considered valuable work. And I think it's going to be very interesting and Hopefully many people will actually take an advantage of that, know, see this as an opportunity rather than a crisis. It could feel like a crisis. I mean, let's be honest, right? People laid off is never a great thing, but it could also be a great opportunity, especially when it's so easy today to build a business. could be a great opportunity to shift how market operates and how people get a chance to actually succeed and fulfill their own dreams, in a way. Yeah, it's dreams.

Niklas:Yeah, I mean, with the layoffs, as you said, the question there is whether AI is the scapegoat or whether it's really the reason. With Meta, I don't know, but these are, like if you have a huge company, every org of 3 % or something of the workforce, that is something that would have happened anyway from time to time. Like that's just the truth. I mean, if you remember Elon coming in at X, for example, he cut I think 80 % of the workforce and the platform is still running. I don't say that that's necessarily the right way, but you can see that these companies, at least in part, are overstuffed, I think, from the past. So I don't know how much there really is about AI or just some shift of R &D resources to some other topic which needs some org changes. And then this really large orgs, I mean,

Davida Ginter:you

Niklas:At the end of the day, that's just a strategic decision to cut some part of the org chart and not invest in a certain area. I don't know how much of it is really just justified there. Meta will definitely try to automate as much as they can.

Davida Ginter:probably both yeah

Niklas:What I was wondering is, do you think training of people will change? Like now we have ⁓ the way juniors get to be seniors ⁓ was quite clear for a very long time. Now we will probably see less junior roles, at least in consulting or services business.

Davida Ginter:Yeah. Yes, actually not just consulting a surface business. This is also true to software, right? If Claude is your junior, you need the person to be the senior. And definitely training is going to change. I noticed recently how many people are chasing to be, even individually, not, know, out as part of the company, chasing to be trained to be AI proficient. And this is by the way, the right call to do. Here's a million dollar idea. Take it. Anyone can listen, can take it to build the next AI Academy. And this is an immediate product market fit because everyone needs it. And training should change, not just for, you know, training people to be a proficient, but also like you mentioned, how do you become from a junior to a senior that can manage orchestrate, ⁓ strategize, because you know if AI takes on a lot of the execution, humans become more in charge of the orchestration of the entire thing. Definitely.

Niklas:you look at the AI workforce and also what you're building, what do you think is the trajectory for the next two to three years? What will happen next?

Davida Ginter:What will happen next is, I believe, first of all, a major improvement of infrastructure because there are still so many gaps there. latency when it comes to voice, inaccuracy and sometimes people call it hallucinations. I don't think this is right word, but inaccuracy and lack of good guardrails or harness when it comes to ⁓ generative AI. So there are still many, gaps that people now put up with because everything is so exciting and new, but in very, very short time, those gaps must be breached. Otherwise people will stop trusting AI to do their work. So this is kind of one area of infrastructure. Another part is how agents interact with the web. Currently agents, and I mentioned this just a few minutes ago, agents just replacing people in searching the web, know. But there's probably going to be, and Salesforce started this with headless websites, there's probably a much more efficient way for agents to interact with the web. So we're to see a bunch of startup building in that direction, either from the agent direction or from the web direction to make this interface work much, much better for agents versus how humans are interacting with the web. And we're probably going to see in this trajectory also a lot of companies doubling down on security because, you know, if now agents need access to pretty much all of your information to do their work, how do you protect all this information? How do you protect data privacy? I think even though I'm already seeing some solution in this direction, this is so early. and not mature yet and not satisfying yet. As a company, I'm concerned. As a company who sell actually AI and AI agents that touch people's tools, this is what I'm dealing with all day long. How do I protect data privacy and security concerns? So this is one of the biggest gaps that will need to be resolved and the sooner the better.

Niklas:Yeah, what I also think is that we will probably see very good Edge AI models in the near future. So currently, I think this is currently one of the biggest challenges that I at least see is we are sending all the data to these RP providers. I mean, we trust them not to store them and not to use that and just process the request. But I think this next large step would be to have local models. And they are getting better, but we are just not there yet. We need too much RAM at this point in time. When somebody now wants to start out in AI and they are bit overwhelmed by the hype, what do you think? Where should somebody start?

Davida Ginter:⁓ That's you know, if I knew the right answer I would be a billionaire. No, I'm kidding I think I think it's still true to what it was a few years ago I mean it seems like okay AI now gets us to let us build everything But think about it the core principles remain the same if you're not solving a real need for people, something that people say, give it to me now, I'm happy to pay for this, this is solving a true need. If you're not building something like that, you're wasting time, tokens and a lot of energy on building the wrong thing, especially because it's so easy to build now. So the best approach has been, always been and still true, is to find a true pain, a true need, a true problem. and solve it in a much much better way than it was solved until now. That's... it's still true.

Niklas:I think these are very good final words. Davida, thank you so much for being on the podcast and to you my listeners, see you next time.

Davida Ginter:Thank you.