From IKEA to AI Agents — Peter Grimvall, Ekona AI CEO

There's a lot of discussion in the industry about it that many predicted that everything is like reaching a certain plateau and we are more and more seeing incremental improvements which maybe has been the thing actually the last couple of months but on the other hand I think we have had just this week Gemini 3 is maybe too early to say but so far what we have been testing it's actually definite workshop but actually what came uh the added day non [snorts] banana 2 that one for us it's actually solved some problem. Hi, my name is Dmitri Bonichi and I'm a content creator, agency owner, and AI enthusiast. You're listening to the AI Agents podcast brought to you by Jot Form and featuring our very own CEO and founder, Idkin Tank. This is the show where artificial intelligence meets innovation, productivity, and the tools shaping the

future of work. Enjoy the show. Hello and welcome back to another episode of the AI Agents Podcast. In this episode, we have Peter Grimald, the founder and co-founder, I should say, of Eona. How you doing today, Peter? >> Yeah, very good. Thanks for inviting me. Great to be here. >> Yeah, thanks for being on the show. Really appreciate it. Uh, just to kind of kick things off, would love to know a little bit more about how you uh first um got into AI and what led you to founding Econo. Yeah, actually how I get got into it was almost um it's almost a decade ago and it was just a coincidence you know I moved countries many times with work and so on worked along in Asia and then um I came back to Europe and then uh after some moves to Germany, Sweden, I

ended up in Switzerland. So the first thing you have to do is like you have to find a place to live and then if your work is far out yeah you need to have a car. So I actually bought a Tesla in a 2016 with autopilot and that actually struck me and uh as I'm working in supply chain I immediately thought is like how can this type of technology actually be applied in uh in supply chain so I actually started to study a bit uh just beside our work at MIT on on this topic and then um I think halfway into the studies I realized that yeah this we have many application areas at my work and as I was already how to say leading that type of topic at IKEA at that time. Um I actually pitched uh to IKEA let's u start the

first AI team and focus on uh on supply chain and u they gave me actually like two headcount so I could hire two people and then we actually started to build some like prediction models to predict the future sales. This was the the first task that we went on and but then I also realized that actually the the same year when I had my parents over for Christmas and you know when you're a teenager you're not so like interested of what your parents are doing but uh then I actually found out that my father was teaching in machine learning like in the 80s but before with power. Yeah, exactly. So I found it out like >> in the 80s. That's crazy. >> Yeah, I think so. 80s or or 90s. um uh now I don't recall no so that is actually the journey and then

we actually just started those um uh use cases and I had the opportunity that I could actually implement these use cases in like my own organizations I could both like develop it and also benefit it uh from myself so I didn't you know I didn't have to go and look for for stakeholders or customers or anything like this so uh yeah that's how I came into And then um to the second part of your question, yeah, how to came into a corner was that um not sure if it was a midlife crisis, but uh after working like 20 years in corporate uh I have a lot of energy and um it would be interesting to try something uh on your own. And what I actually discovered when I venture into to start something on my own was actually the combination between actually actually knowing something

plus knowing a bit something a bit about AI is actually very nice uh combination. So I worked 20 years in uh in supply chain and everything from procurement to manufacturing and logistics and also like design and and planning uh etc. And then also being early on to IKEA. So that meant that when we are actually talking to to our clients and working our projects, we can actually connect very quickly. Uh and also I think we can also avoid many of the simple mistakes when you actually don't fully understand the context uh of why we are building something. So u that's when I got a call then from um my co-founder that has actually like more or less identical background but from pharma instead we said that okay pharma supply chain and then I have also worked a bit in finance the last year so we

said okay pharma supply chain and um and um and finance that makes a good mix and which is also a bit like the core industries here in in Switzerland where we are based also Okay, very cool. Nice. Um, you know, I think that that gives us a good kind of like starting point with with your whole story. So, what was kind of the [snorts] the turning point that made you realize there was a market need for um what you were doing? I think it was actually when we started to meet clients and when you have this connection and when clients are actually ready to shake hand almost like on the first or at most like the second meeting. So let's go you know more or less when you know when people are ready to shake hand without knowing all the details. So we have established

some type of trust. Uh and then we thought okay this is actually a very nice niche to to work in and it's also very fun because it's sort of you know I enjoyed working in supply chain so I sort of you know I stay working in supply chain but I also uh work in AI. >> Gotcha. Yeah. No totally I I agree. I think that's um that's definitely uh yeah I I I think it's interesting kind of where this uh intersection of of everything plays out cuz you know when it comes to your journey um just to kind of get more in deep after the beginning of it AI is obviously a great uh timesaver it's a great um expander of cap capacity right um according to some of the research we did you guys managed to have really solid revenue and growth only with

like a 15 person team, you know, what does it kind of look like to to be uh a company that's AI first, obviously an AI product yourself and kind of manage to run such a lean uh tight ship um with uh what you're doing with AI and and what what's that experience been like? Yeah, I think first of all is um we have been super lucky with the with the timing because as you say to be AI first has more or less only been possible very very recently. Sort of means that everyone that started like 10 years ago they sit with the lot of legacy and probably a lot of like legacy competence and and resources in the organization. Yeah, we were actually able to to start very recent and then we can actually be AI first and actually having the approach on building products

products with AI but also developed with the help of AI. So um now I think it's actually more or less pure luck that uh we we got the the right timing because today you can do so much. Uh >> a man makes his own look sir. I'm just kidding. No, but I think it's like the the huge difference is I think if I just go back maybe like two or three years ago then maybe you could focus only on AI. You will do the AI part but you will be super dependent of having like another agency or another team doing the UI and the front end and so on. Now I sort of do the front end and it goes quite uh quick. Yeah. And uh and as I also as we have the industry experience is that to create the front end when you

sit with the experience without having all this translation to like a UI expert uh software engineer etc. you can move very fast. So I think you know the first like real product um we built at IKEA you know to do like a demand sensing system and roll that out globally to 400 stores. >> Sure. Yeah. >> Maybe that took like two years or something. Yeah. Now we are talking maybe 2 3 months to build something similar. It's crazy >> and also with much fewer people also. >> No, no, that's that's totally fair. And I think it's interesting uh from from my perspective because like you do have like a 15 person team, right? Is that >> Yeah. No, we are we are about 20 people but >> Gotcha. Okay. Um, what do you what do you think about like the ability for what you're doing

um to be kind of I guess I don't want to say enhanced um via having like a smaller team because for me it seems like the more lean you are and the more agile you are especially in this world of AI, you can really build the foundation with these AI tools a lot better as a company. Um and and kind of following up on that, I would love to hear more about like how that kind of fits into some of the philosophy that you have um for your practical application um into companies uh workflows themselves too. >> Yeah. No, I think uh if you go back a decade or even just 5 years, I think you know you will win a lot of clients saying hey we are like 10,000 people or we are 20,000 people. We are very capable of doing things. Now I

almost think that it's actually not helping these companies to be 10,000 or 20 TCD. Yeah. They're simply too big, too slow. So I think actually speed is actually a more strategic asset than than size nowadays. Yeah. Um so um I think that is actually what we can say that we can work fast and we can be uh close to the development both from a submaster expertise but also from like an AI uh expertise and then um yeah we can have quick and interesting developments. No, totally. I think that's that's very fair because like practically speaking, um when you get to the size of a 20,000, 30,000 person company, you know, I mean, I I I've been at decently sized companies of a couple hundred, 400, 500 people, even at that stage, there's this like funny level of red tape that occurs. You're like, I just

don't know why, right? But the bigger you get, >> the worse it gets. um the more approvals you have to go through, the more layers you have to go through where it's like if someone just implemented this without the quote risk, um I think it would just go quicker and better. >> Yeah, I think a good example is uh up in the Nordics uh where I come from is actually this is Lovable. I don't I don't know how many they are, but I think they're Yeah, they're I think they're just like 40 people. >> No, shoot. I love Lovable. Big fan. >> And they are No, a super big company. at least the revenue but not uh in terms of size and employees. Yeah. So >> I think they're on track for something in the realm of like 200 million ARR. >> Yeah, I think

I read uh something >> it was like came out two days ago. Yeah, >> I had uh the opportunity I think it was more or less uh one year ago. Uh I think it was one of the co-founders actually demoed the product for me, but then it was like super immature uh one year ago. So, but it was I really hope that you know just from that stage >> to go to this uh big type of uh spread around the world. >> No, absolutely. I totally agree. That's that's that's really cool. Um that by the way that that he was able to chat with you about that and show you that. So, kudos. Very cool. Um what I would say just kind of taking a look at it is um [snorts] you know you are kind of in an interesting position. I actually have a

a question for you about, you know, what you guys do just to dive into a little bit more. What would you say um AANA solves for your customers um mainly and what are those types of industries? I know there was a little bit of pharma reference and other regulated industries that were mentioned on your website. >> Yeah. Well, that is one thing that is uh important that you pick is like regulated industry actually tend to be a bit also of our specialtity because uh it's not only to develop the the solution it to develop also the overall like uh like governance. So like for example in pharma you deal with a lot of patients data uh and that you need to have a very rigorous governance and security around finance is more or less the same. It's just it's not about the the weight and

the diseases of a person if the credit card details of a person and that you have to take equally uh good care of you. So um so I think that is an is where we're working on. Um but you can also divide in like in the technology I would say still maybe onethird of our work is more of this what I would refer to this uh traditional machine learning where we build like forecasting models who are predicting different events. Uh we have also been uh active in account frauds and so on which is also traditional uh machine learning to the majority even though we are moving in a bit to hybrid in both of them. So for example, what I mean with hybrid is like maybe we use gen AI to qualify some signals that can be used in features for machine learning models for

example. So for example, if you want to digest a lot of like social media signals, you can leverage Gen AI to actually digest that into something that you can use as a feature in the like a forecasting model. But then the rest two/3s it's u it's purely gen AI and of course a lot of agentic solutions and >> but maybe you want some examples also then what we do or >> yeah no absolutely I think that would be that would be a great yeah follow would be what are some examples >> we do uh a few so I think in if we take supply chain to start with we mentioned like the forecasting and so on and then if we move into more the the aentic AI it's a lot of workflow automation and that could be anything from like how you manage orders, how

you manage invoices, uh, etc. So, typically where you already have a group of like five, 10 or 20 people doing a little bit the same thing. So, we have like a defined process and we are actually more or less uh automating this type of process. And it's also that when you have those larger groups of people, I think it's also easier for our clients to harvest the the savings also. uh so when they are thinking about when people are leaving they can actually think okay do I need to replace or uh can we manage and scale better with the with AI solutions so that is one thing uh >> and in pharma it's a lot around uh there's a lot of work around compliance and there AI is actually very very good there's also a bit of workflow compliance is typically you have like medical

compliance regulatory compliance maybe legal compliance. So for example, what are you allowed to say about this product? How are you? So you make some typical like medical claims that you either use in your marketing, you use them in your packaging, etc. So that is something that all like pharma companies uh medical device companies and also a bit more on like consumer health companies uh are dealing with on a daily basis. Yeah. So typically like a very long troublesome process where we can help with automation but also like pre-qualify. So that means like avoid submitting material that will not pass. Yeah. So that is the worst thing. Yeah. You actually you create something but it will not pass but it will you know it will go through some of the reviews and then it will get stuck on the last review. So with AI we can

help them to be like pre-qualified to a very high proability. So that means that this flow instead of taking like several weeks or several months can now run in a couple of days through the organization. So that's a very interesting use case. Uh then another case that my co-founder has worked a lot in uh is also how to leverage like clinical studies. So you have done studies in the past but now you actually want to look at like selection u so you have registered of course a lot of of data in this study but now you want simply actually more or less ask another question. So in the past it was you know there was someone that had a question but could not code. So this person had to go to like a data scientist and then this data scientist you know retrieve the data

uh clean the data organize it and so on and then deliver the insight back. Yeah and that is something that we are actually building a lot independent of industry is more or less this text to SQL. So with natural language you can ask questions to your data and not only get the answers back but also maybe with visualization maybe also some proof points also so you actually believe in the answer. I think that is also always a challenge for example if you use the just chat GPD you just get an answer but you know no reference. Hm. Yeah. No, that makes that makes a lot of sense to me. I think um it's it's interesting, you know, kind of where this sits because I would imagine and call me maybe correct me on this a little bit, but um most companies that are in regulated

industries are on the surface you'd think a little bit more apprehensive towards or cautious about implementing AI solutions uh right now because of security reasons. Is that a fair assessment? >> I think we can say we are experienced like uh two sides of this um pharma. They are very rigorous about the data security but they're also very pro but they're also very progressive >> like they want to do it but they got >> they really want to do it and they are doing it. when we work in finance >> many of them are still operating like purely on prem uh which is interesting today yeah with the open source model and uh the open AI latest models for example so last week uh we were in uh at a meeting with one of the big banks around here one of the largest in uh in

Europe and we shut off the Wi-Fi and we demoed what and like uh just what we had on the MacBook what you can actually do today. Yeah. So this is quite uh amazing development. So that I'm actually very interested to to follow you know maybe in the future we will not ship a product maybe I'll ship my my MacBook uh pre-installed with the software and then you can work offline if you want the ultimate security but I think in the end uh also financial institutions they are moving into the cloud. Yeah. But then you can imagine it's in Europe is also important that it should be like a European cloud. The data should stay in Europe and not go anywhere else etc. So there are many parameters to think about. >> Yeah totally. Um there are a lot of there are a lot of parameters

to think about with this and and you know I actually my dad works in finance uh in the states and I think they're they're in a very funny position especially when it comes to security like uh anything I feel like that comes up in finance is regulated at this point. um anything and everything is uh very um regulated. So I would say you know another question that is important to a lot of people is you work you know across building you work across building multi- aent workflows right I think now we've kind of learned about orchestration agent sorry agents that orchestrate um there's a lot of infrastructure and strategy questions how do you kind of decide which part of um that uh journey each client needs whether they just need some individual agents how do you determine and how to build multi multi- aent workflows.

What's your like approach for that? >> Yeah, actually it starts a lot with um actually if you want to really zoom out, I would say it's like it's actually no difference with AI compared to when you actually maybe 10 or 20 years ago wanted to digitalize your your process in some way. It's like you really have to understand the process. You have to make sure that actually the the process is also like followed uh to a high degree and then uh you also then have to break down the process. Yeah. Because as you know you are you are also working on this on a daily basis. If you write this like long prompt uh that should cover you know many many steps on the process you know you're likely to be correct. is not so high. But if you break down the process and uh

tell the agent to focus only on this small part, you have a much higher likelihood to to succeed. Uh so that is also of course a big uh work and then you talked about like orchestration and so on. So there is always a question if the clients already have some infrastructure in place. I would say when it comes to maybe aic AI most of our giants don't have anything in place. So then we typically build also like the observability so we can observe the performance over time of these agents and then uh it's also the traceability is also very important in for example in finance that you can actually trace back every single prompt that those agents has done because if you are in like a compliance case you need to be able to prove actually you know what actually happened what did this agent

conclude and ship to the other agent etc. So you can actually trace it back and actually we use it also for problem solving and and bug fixing also. So it's not only for like compliance actually we use it a lot for for our own development also. >> Yeah. Because then you can kind of see where things uh um can be not only just like you're saying it's it's not only always about compliance. Now compliance is definitely important, but it's definitely important to kind of note what um what improvements you can make and why where it went off the rails and why it went off the rails. So yeah, a a minor question I do have for you because you know working in this realm I think there's maybe some interesting thoughts and misconceptions that one could have about like models and how they impact things.

I'm not saying that you know models mean nothing. Obviously they mean a lot. Um, but at what what I'm kind of asking recently a lot of people like you is at what level have you noticed like the biggest changes in model capability impact things practically? And then like kind of where has maybe there been like a bit of diminishing return because I think like 4.5 sonnet was amazing. I think Gemini 3 was amazing. It just came out. >> Yeah. >> Um but like 03 you know was like such a big step in reasoning. Um, and I think reasoning models were big. Uh, but kind of a in the last couple months, I've sort of had this feeling where we're seeing marginal improvement in a lot of things. We're we're maybe not seeing the same massive kind of leaps we were seeing before. Um, from a

practical standpoint, maybe coding agents are going to continue to kind of like blow our minds, but where is this kind of improvement in models helped you guys out? And how have you kind of felt that they've they've helped you out a lot? and more maybe have they kind of only helped you out uh in sort of a marginal sense because that's that's the things I'm asking some founders and I've getting a lot of interesting different answers recently. >> Yeah. No, but it is a very interesting topic. Yeah. Because it's a lot of discussion in the industry about it that many predicted that they everything is like reaching a certain plateau and we are more and more seeing incremental improvements which maybe has been the thing actually the last couple of months. But on the other hand, I think we have had just this week a

bit of I think uh Gemini 3 is maybe too early to say, but so far what we have been testing, it's actually definitely not sharp, but actually what came uh the other day Nano Banana 2 that I was like that one for us, it's actually solved some problem. It was like >> like what? So uh coming back to uh when we are generating like marketing materials um so I think Nano Banana was really good in generating like draft material with the visualization and everything but was not so good in accuracy on the text. So they will put the text but a lot of like obvious spelling mistakes. Yeah. No banana too. It seems like okay we have just you know tested it for like 24 hours but uh it seems like a majority of these things are are actually gone. So now you because a

marketing material is typically you know you maybe have some background or like a person you have some illustrations of the things that you want to promote or show how how good things uh are and then you typically have text also. Yeah. And um if you want to build tools that are generating this for the for the creators and so one of our marketing campaigns then the combination of all these three are important and I would say Nana Banana was good at two of them but now Nana Banana 2 actually does all three of them. So >> Nana Banana Nana Banana Nana Banana say that fast um >> yeah exactly >> 10 times. It's rid it's actually ridiculous. No, but I I I had noticed that it was good um in Nano Banana 1. Um I haven't uh checked out two um too recently. So, I

do appreciate you calling that out and I don't know how the heck I missed that, but um and am excited to kind of test it out after this episode ends. Yeah, but do that because it's also like it can also I haven't tried it much but actually the examples I got from the team today because they put like a lot of like screenshots in our common chats today and it's actually can understand very much the the context very well. So I think uh one of one of the team actually just linked their like LinkedIn profile and then Nana Banana did like an super nice illustration of this person's career step by step what they have done and so on. Yeah. With illustration, the year, the text, everything. >> Super cool. >> Uh it's funny. Uh and it's available I think right now in like uh

API if I'm not wrong. Right. It's not really like Okay. Yeah, that's fair. Okay, so that's a good note for me. Thank you for that. And everyone, make sure to go check that out because it's pretty dope. Um, speaking of kind of the improvements that we're seeing in AI, right, and the capabilities that it has. Um, you know, what are some of the really cool um things that you think will come out of this from a job standpoint? Because I think a lot of people are definitely concerned um or really excited about what it means for jobs. And um I kind of want to get an opinion of yours to uh you know understand where you stand on the whole is it going to create jobs? Is it going to remove jobs? Is it going to make jobs just different? Where do you stand on

this whole thing? >> Yeah. Hopefully it does both. Yeah. Uh otherwise it will be boring. But I I have actually faced this question the last u 10 years. Um and most likely people that worked before me uh at the companies I worked for that worked with the digitalization and so on they probably faced the same question for another decade earlier than that. Yeah. Um and what did I learn? I think if I look back it's like can you mention a profession that existed you know 50 60 years ago and doesn't exist today. It's not that many. Yeah. >> Yeah. Not many. Yeah. >> There are not many. Yeah. But I think there is one that is totally gone and that is like the elevator operator. That is very very seldom you see uh today. So that has been replaced by buttons with the number on

and people can push uh themselves. But otherwise it's like you know you're replacing 10% you're replacing 50% 80% 90% of a job. So of course if you have a big mass of people doing the same thing you will find some um some improvements there but otherwise it's still very difficult to fully replace a job. Yeah. You know even autonomous driving today is based a lot on having people in the back office. Of course, yeah, one person can deal with, you know, several vehicles at the same time, but you know, it's not that one person totally disappears. And then you should also add in the people that it takes to develop this and uh and maintain this, etc. So, I think um yeah, I don't know if I want to give a number, but uh maybe it's plus minus zero. >> Plus, minus zero. That's funny.

You know, fair. Um, I think uh the elevator operator is a good example I might use moving forward cuz it's like it kind of illustrates the lack of like who cares, right? Like it's like okay, there's no elevator operators. Who cares at this point, right? Like it's it was such a very specific need that no longer obviously is needed, right? It's it's it's so abundantly clear that it is I'm trying to I've been trying to find a way to phrase it. Maybe you can help me. There's a difference between value additive uh jobs um and then there's a difference between I don't want to call it keeping up with the Joneses because that's not the right phrase but basically like jobs that are functionally able to do a task that keeps the thing going but it doesn't require any actual like mental uh or physical

prowess right like I don't know what the word is like cog maybe just cog jobs like people are people always talk about it's like oh I don't want to be a cog in the machine you know in like the modern workplace um but then when maybe those types of jobs would be removed by AI they often are like afraid of it and I find it a very like a very funny um cognitive dissonance that exists there >> yeah maybe those that are like more cog in the wheel I would say they are more likely to be replaced by robots without AI because if you're doing exactly the same thing over and over again. You can have a like a robot and I think if you go to any modern manufacturing plant today, you know, there are >> Yeah, that happened already of of robots. Yeah.

Yeah. >> So, uh absolutely >> that has already happened and I think it's like it has not been like a financial crisis because we have introduced robots in our manufacturing. >> Yeah, that's a good point. People don't really talk about People don't really talk about that. Yeah, I think there Well, there was a People don't really talk about the fact there was an entire move to um elevator operator. There was uh well, that's just the smaller bunch of people, but the main people being um manufacturing work I think was decreased by these robots you're mentioning. And to be fair, you know, obviously it did mean something, but practically I do think people adjusted you know companies adjusted um we've continued to see economic growth and people taking advantage of it so yeah it's a good that's a good and interesting point so >> but you

can also take another angle saying that but what will happen with companies that does not adopt AI yeah there is no guarantee for survival >> well that's that's on them I'm just kidding >> no it is kind of on them to a certain extent because I think practically you know we have a lot of opportunity here to continue to, you know, do what we're we're doing and it really positively impact people across the board and sometimes it's good to force people to get into like a higher level of uh you know knowledge work and let's say for example that's not their stick. I do think there might be some level of like personability that comes back to work in an ironic way when we remove all of the non as heady knowledge work. >> We do a lot of like learning a craft. Uh it's

like u I don't know what I should recommend my my kids to do but it's like learning a craft and that might be a very valuable skill in the future. >> No, I totally agree. Right. like physical crafts I think could come back because there will be almost it'll be good knowledge work might be so commoditized um like basic knowledge work I mean I I think still like premium like strategy stuff will probably be like human still because people can come up with stuff in their head that I just don't think pattern recognition can ever get to um >> didn't what's a guy Jeffrey Hinton didn't he when he won the the Nobel Prize last year in an interview he was asked and then he recommend and that you should become a plumber cuz that is very very difficult to replace with AI because you

will work in very like uncomfortable positions and you will work with your hands and every house that needs to replace the water pipes and sewage pipes. It all looks different in every single house. So >> yeah, I think that's a fair that's a fair assessment. like there's there's some definite difficulty um that's in a lot of these jobs that people don't appreciate. And like probably one of the most lucrative things you can do right now would be taking a company that's more brick and mortar andor more bluecollar. Um this is apparently something people that are like in their 30s with some capital are doing. They're like buying brickandmortar jobs or brickandmortar companies or bluecollar companies like uh plumbing companies and they're automating a lot of the work that's more knowledge work. Um they're optimizing routes and from there all that together is giving them like

a much more like high profit margin bluecollar business and they're making great money with it. Right? So there's a lot of opportunity in that sense that I think people are maybe unaware of. And obviously buying the company is like not on the table for a lot of people. But you could build a company and a subcontract a bunch of different plumbers and optimize it better than like a older um owned uh company would be, right? Or um yeah. So that's something definitely I think that that could change in the future. What do you think is um the most misunderstood thing about AI and business right now? It's a good but if you take really the business perspective I think the most misunderstood thing is when I hear people I use chat GPT at work because then I don't feel that they have fully discovered maybe

the full potential. Uh >> I think that if they just respond with that I don't think they'd use AI much. No, exactly. Uh, yeah, that's maybe true. It's a maybe quick answer, but I think it's like to use very generic tools in your work. I don't think that will make you the the market leader so to say. I think you should more think that with AI you can actually build something like that is super relevant for your business today. Yeah. And that is the >> I totally agree. I totally agree with that. I think my favorite thing to to ask people is what's their favorite uh AI model. >> And if somebody can't answer the question, that means they don't know Jack about AI, right? It's a very easy veter. Like I'm not saying they'll know everything if they tell you Claude or if they

tell you, you know, Gemini 3.0 was a great release today, right? But point is I ask that question because it tells me that they I they know nothing or something. >> Y >> yeah. Um beyond the um you know basic hype of what's going on, right? You have your your Geminis, you have your Nana Banana, um you have so many different tools out there. What do you think is the most valuable thing practically on a daily basis that companies can implement with AI right now? >> Yeah, I think most companies uh have it but I think just uh yeah to have a very good search. I think that is uh is something uh like I think in in bigger corporations you tend to have like a lot of documents. So to have a a really good search I think it's uh like you know but

connected of course uh like uh with an agent or like an LLM so like a rag solution of uh of any kind. So I think that is something >> this is a pretty good answer. I that's a pretty good answer. I didn't consider that. So you think it's more um >> if you have nothing if you have nothing. >> Yeah. No, I would at least start something because I think it it's it's in a how to say it collects points in a in a certain dimension. First of all, it's a very simple thing to put in place. So, the investment is extremely low. You know, a couple of days work and you will have it uh up and running. It's also a bit like democratizing because I think today you actually if you're running a company you want a bit about the awareness because probably

the the good ideas will come from your organization. Yeah. Um because there there are a lot of shitty tasks that are being done very repetitive and so on. Yeah. Might not be so easy for the for the seuite to know all this. Yeah. But actually to enable the entire organization to get some knowledge on what is the art of the possible you know then you will actually get the all ideas to to improve uh and so on. I think that will be an easy easy starting point and you can also imagine you know how long time people spend on you know where do I find this document uh what did we agree one year ago what did we say at this meeting or something like this. Yeah. So that I think is an easy thing. But otherwise what we typically do is uh we ask

what about what is the thing that you like the least about your job. That can also be a very good starting point on on finding opportunities. >> That's a very good starting point in my opinion. I I think that's a very good point is just like there is so much that people dislike that they can get rid of. I had a conversation with a with a friend of mine a couple days ago who was like, "I heard you're into AI and stuff and um like asked me questions about what can be done." And it was so funny. They were just focused on this very minute uh in uh setting up a new client, like just being able to like have a templated Google Drive that has a bunch of PDFs. I'm like, "Bro, this [snorts] is possible in like 2017, right?" Like this is not

like AI. Um I'm like Integramat and Zapier has been around for a while sort of sort of type conversation. But then when I explained to her that you could actually like create agents that are able to analyze floor plans and like make assessments of how much to charge for the drywall work and like have that be an auto response basically is I was like you had to spell it out for me. You're like how long does it take for someone to make a floor plan? I don't know like four or five hours. Okay. Is it basically square footage and like analysis of how many walls and stuff? Yes. I'm like, okay. So, it's pretty much like if this then that. Yes. I'm like, okay. AI can pretty much do anything with reason that has if this then that logic. So, you how many proposals do

you do a year? How many do you do a year? I think we do like two 300 300 a year. I'm like, okay. So, the average person works 40 hours a week, right? about 48 weeks a year. I'm like, "So that's 19 thou?" I was like, "That's 2,000 hours. And if you tell me to do like 200 300 of those, you just replaced a third of somebody's job." And they're like, "What?" And it's crazy. People are not at that level yet. They don't like it's it's not very well known. I don't think that that's like how easy it is to find stuff. So, if you had to give like one recommendation for a personal tool that saves you time, I I always like asking this question to kind of close things off and that'll let you uh plug what you're doing to to end the

>> Yeah. No, I think a personal tool is um what I like is uh so for example, if you use uh just the normal chat GPT like the app actually what I appreciate a lot is the memory it actually you know remembers because while listening I work on something in in my head and then maybe I use like um for example then chatd I then I I'll use the and then next weekend I have some more time to think about the same problem or I have discovered something new during the week and I want to add that into it etc. So if I just take a a very simple example, I uh during COVID, you know, I was working more or less 24/7 because supply chain and co was not a good combo. But it was actually fun fun times working, but it meant a

lot of work, sitting still a lot. So um after one year, you know, sitting on a chair, I started uh running and then this evolved into trail running. And I think this year I've done like four or maybe yeah four races in trail running and I'm starting to get better and better and shy is like perfect you know I log my results there and then I say okay you know in two months I have this race and the trail running is typically like long and it has a lot of elevation. So it's not like if you just run flat, you know, it's easy to compare, but now you have to compare a bit like how technical the terrain is, how long the distance is, and how many how much elevation I need to climb in the race. Yeah. And it's perfect because it remembers those

and then then it can more or less tell me that, okay, if you run this race in two months, you will finish around four and a half uh hours. Yeah. Um and then when I've done another race, I can add that. So if I you know if I'm improving then I say okay you know the estimation I did for this race now we can actually reduce that by 10 or 15 meters. So the memory I think for me is very useful. >> Yeah that's [snorts] a very good point. You know I honestly dude the the best thing for me um in my opinion is um I'm recently using so I like cloud a lot. That's like my favorite model right now. I think Gemini 3.0 is really good too so it's kind of hard to know. I like text core text. It's a tool that

allows you to like make agents and personas and um have models that you can switch between as like web search and all that kind of stuff. So, it kind of like set the infrastructure for I know it seems like a lot of those tools might not be useful where you just like it's just like a model um hub, but this one does a really good job of having like really good writing agents and stuff. So, that's good. Um, and there's another tool that I recommend to all business owners and people are on the go with meetings. It's uh called Crisp AI. So, it's a KR ISP AI. Um, it's a um it's the first like meeting notetaker I've seen that like really has noise cancellation. So, say you have you're on a train, you're in your car, whatever it is, and you're taking a meeting.

It has about like a half second or full second delay from when you speak, but it like actually noise cancels the background noise. >> So, okay. Yeah, maybe I'll try it out. We actually built maybe next time I'll demo that for you. Like um actually a video editing tool, but more for scientific um videos because videos are used a lot for for training materials >> for example uh like for specialist doctors and so on. So uh there we actually built a tool that not only that it transcribes and it also keeps like the balance uh because uh like in far as you say it's very important to be correct the regulatory. So if you have several speakers several doctors uh speaking about the topic. So if they are speaking for one hour and then of course the specialist doctors they are always on the run.

So they will like have a format that is five minutes and then we can edit down the video but we keep the same ratio. So if Dr. A and Dr. B spoke 50/50 they also speak 50/50 in that and then we use agents to do the fully correct translation by for like example uploading the scientific papers so you actually know exactly the drug names and so on because otherwise it could be like blah blah blah blah on something very crucial. Yeah, because those um trans um translation models they are very or transcription models they are very generic. Yeah. So they work for the general but when it's specialist talks >> could be something >> true. No that that that's very interesting. I I think I've been looking into the world of AI and um how it kind of works from uh the standpoint of like

uh video editing and where that can lead because I don't think that's actually something on the market right now that's well done outside of just repurposing it for like video shorts. Um I think it's kind of a a a spot in the market a lot of people would probably benefit from. So >> yeah, I'll send you I'll send you a link. You can try it out. >> I'd love that. I'd love that. Um, and with that being said, Peter, uh, where can everyone go to find you to check, uh, you guys out? >> Yeah, so you see this, uh, here we are in the center of Basel in Switzerland. So, next to the train station, next to Maral, which is like a a big foot court uh, here I think you knock on the door. Actually, today we had someone knocking on the door a

couple of weeks ago. Actually, we had like just a client knocking on the door saying, "Hey, do you work with AI?" Um but otherwise uh our website uh econom so e ko n a uh I think is the easiest way to reach out to us. Actually I didn't tell you the story either by about the name. Yeah. So I think there are many like companies that are coming and going in the AI world and I think we wanted to make it a bit more like longterm. So uh and is also that we are not like a uh venture back. So we are like a fully private company with with two co-founders. So it's actually the initials of our u wives and and kids that makes up a corner. >> Yeah, that's very nice. I wouldn't have guessed that. That's very cool. >> No, exactly. >>

Very sweet. I think uh it's always good. You got to give them a those are the people that make you do what you do, right? >> Yep. Exactly. >> Get you out of the bed in the morning. you got to grind as a as a founder. That's awesome, dude. That's that's very sweet. >> Small kids take this job very seriously. Get you out in the morning. >> Yes. Yeah. Yeah. Yeah. Yeah. Right. Well, with that being said, everyone, please make sure to go check out what Peter's doing over there at Econ. I think it's um that's a really sweet way to end the episode. And um with that being said, also leave a like, subscribe, comment, and do everything that you can to help support um not only our show, but what they're doing over there at Eona. Thank you so much for watching this

episode and we'll see you guys in the next one. >> Yep. Thank you. Thank you.