52: Using AI for Smarter Enterprise Solutions with Edwin Lisowski

This podcast episode is sponsored by chiefaiofficer.com offering training and certification through the International association of Chief AI Officers. Interested in a new career or leveling up your value in the marketplace? Chiefaiofficer.com can help.

The following is an interview with Edwin Lasowski. Edwin is the co founder of Adepto, a top AI consultancy recognized by Forbes alongside industry leaders like Deloitte. Since 2017, Adepto has been delivering cutting edge AI solutions across industries, from natural language processing to generative AI. In this episode, we explore the evolution of AI adoption, how businesses can use AI agents to solve real world problems, and the steps companies large and small can take to successfully integrate AI into their operations. We also dive into Edwin's perspective on the future of AI, including its potential to enhance decision making and drive innovation.

And let's begin.

Welcome to Using AI at Work.

I'm your host, Chris Daigle.

Each week we'll be learning how today's.

Business owners, entrepreneurs and ambitious professionals are getting more done with smart use of tomorrow's tech.

Let's get started.

All right everybody, on today's episode, as I mentioned, we have Edwin Lazowski, the co founder of Adepto and doing some fascinating things and has been in the space for a while phoning in from Warsaw today for this. So thank you for being on the podcast today. Edwin, tell everybody a little bit about kind of your journey to where you are today.

Of course. Cool. So Chris, first of all, thank you for inviting pleasure to be here as AI is a close topic to me and to my company. Yes. So we are with AI for a while. I mean we started company in 2017 and since then we were deploying different AI solutions to our clients. So as you also, Chris, noticed, we've been a while before ChatGPT was even introduced into the market. So we see how this technology was emerging for all the last years and how basically that being adopted today as well. Yeah.

So how long have the operations been, I guess in Warsaw and New York? When did you open your New York office?

I mean, since beginning. Since beginning we've partnered with some people who are based in United States. So our, our goal was basically, you know, bringing our talent from Europe to United States overseas so we can collaborate with, you know, recognized brands and big, big companies in, in US And I.

Noticed that you guys have been working with some, some very large companies and, and an article I saw in Forbes from again before GPT 3.5 was released, you guys were on a list with some very heavy hitters like Deloitte. On the top 10 AI consultancies. So how did that happen? And that's a quite an accomplishment.

I mean we see that also the work that small companies as Adapto are doing can be also recognized. Yes, nice. Yeah. And the one thing I would like to mention is we usually in some tenders or you know, public RFPs or the project tenders, we compete with such large players. Yes, nice. And Adapto in some cases is selected because of its, you know, expertise, agility as well as, you know, how we approach topics in a very fast manner as we are small, agile and very fast company. And we usually, you know, bring this approach and bring our, you know, flexibility into our clients projects as well.

Are there any like particular industries or categories of product that you guys like to focus on?

I would say yes and no. We've been partnering with very different industries, starting for example from real estate, going through retail, manufacturing, entertainment industries, but also partnering with such interesting industries like aviation, airlines, aerospace. So it's really, really, you know, very different verticals and industries. Yeah, great.

So as I mentioned before we got on the phone, my particular interest and I think the gateway drug for a lot of people into AI is the generative side of things. Now you guys have been doing way more advanced stuff than the average user for sure. And what I'm particularly interested in is when did you start to pay attention that generative? First off, did you have any idea that Generative AI's application would be adopted? So to me it seems like it's everywhere. But I know that we kind of live in a bubble of the AI space and also like what, what. When did you first notice that this is something that we really need to pay attention to people who are in.

The industry for some time we were aware of companies such as OpenAI. Before introducing ChatGPT, OpenAI had some kind of libraries I could say that we were using in our development or projects they call just GPT, you know. Yeah. And ChatGPT is an interface to. To. To like to the product or project we were building for years before introducing in 2022. Yeah, we as a company paid attention to the generative things. Again I think since the company's beginnings as even then we were like it called a little bit different. Yes, we recalled such projects. Natural language processing project, Natural language generation project. And in. In those solutions weren't so smart as ChatGPT today. But this solution solve very concrete small problem related to text or natural language. Yeah, yeah. Well even before that we were like, you know, aware of such technology but like really the perceptions of what is possible with this technology changed when the ChatGPT was introduced in the to market.

Okay, this is interesting. So me, I'm a non technologist and as a matter of fact when ChatGPT was released I saw, I, I knew instantly, wow, huge opportunity. But not for me because I don't know technology. Right. Like I'm more of a business, business operations, business growth and scale. So the initially I was a little, I wasn't that interested honestly. But you guys, I guess you were there before even the GPT paper came out at Google and you guys were already in AI. So you've been really witnessing this from an experiment, a theory to an experiment to a more interesting experiment to really something that's established itself let's say at this point, right?

Yeah, yeah, yeah, that's accurate. And I mean this is how things were moving. I mean as you mentioned like the Google OpenAI and all the large companies were working on such technologies for years. Yeah. And we were just using that to solve like minor problems or smaller problems. Today with this technology that already introduced into the market, we are capable to solve bigger, larger scale problems.

We'll talk about the actual impacts of what that looks like in before you were able to use these smarter versions of these tools and after. But I'm interested, has your client base changed at all since there's now this, this kind of more accessible version of AI?

I would say not really the only thing we've noticed that as I said like people, some people were not believing in AI before ChatGPT. I mean yeah, it was kinda okay, AI, this is maybe statistics, maybe some rules, maybe some coding, software engineering plus data that that was AI, you know before, before ChatGPT and all of that was about you know, predictive analytics, just predicting some things like demand forecasting, you know, churn prediction, computer vision stuff, you know like detecting objects, people faces, recognizing some, some images and moving to the text processing chat GPT showed to the whole world basically how powerful is technology. Yes. And we were capable to do a lot before ChatGPT was introduced. But challenge was also to you know, make people believing in it. But now people become more creative when it comes to AI because they seen like in real life applications or actions how ChatGPT is powerful.

Yeah, as you were saying that I was thinking about that prior to like what happens if you're not a technologist, you're just a business professional going about your day and you're not involved in the architecture, the application of the technical side. It's A black box. And you don't even understand that it just gave you this business intelligence. Right. But when it's in your face, when it's on your phone, when it's in your car, when it's on your screen and anyone can say, hey, I'd like to build an app that does blank. And the tech, not that simple interface allows the technology to. Again, not saying that they're on parity with a PhD in computer engineering, however, that has certainly expanded the awareness in the general public about this quote unquote artificial intelligence. People have all kinds of ideas about what that means and what it means for the future and that sort of thing. So it's fascinating. We don't normally get people with as strong of a technical background as you on the podcast and I have a lot of questions about that because my position in business at least is that for most businesses that, that that are hearing about AI, machine learning isn't the top of their list. It's hey, how do we do this faster with fewer resources or whatever. So a lot of times I, I don't even pay much attention to the technical side because I know if they need that, we've got resources. Hell, now we'll call Adepto and you know, send you guys if you want the business. But this is interesting. So the clients that you already have, I would imagine that, that they've been gaining benefit obviously from the work that you've done. But now that they're seeing these things, is there a lot of repeat business? Are they saying now we want to introduce it into operations or now we want to introduce it into product design or anything like that?

Yes, I would say yes because when we see what is capable within again chatgpt, you can think of a lot of use cases, but still in a limited way because you're limiting to the possibilities of what you already tried used and you can imagine, okay, what kind of business problems I can solve with such tools. So today use cases are very repetitive. Yeah. But with some unique solutions for some industries or departments. Yes. And let just give you an example. After Genai was introduced, all the companies started building internal knowledge based chatbots. Yeah. So everyone will start asking questions about the company about compliance process procedures.

Except for internal use employees using accessing internal knowledge.

Yeah, yeah, yeah. But now it's starting, you know, expanding, evolving a little bit and we starting adopting that to a niche specific, you know, use cases or business problems like, okay, what if AI can read our internal knowledge, internal database documentation and what if this documentation is way complex like it's engineering documentation with the code. Maybe some graph architecture, maybe some technical design. What if we can using that data, start generating more complex information like documentation. Engineering documentation, simulations, documentation. Yeah. What if we can start drawing architectures or very complex technical details? Using AI? Yeah. Those are very. I would say more complex use cases. Those use cases take more time to implement. But those use cases bring a real business value. Yeah. Not just, you know, having and chatting with my company. Documents. Yeah. Bring more real value and automate complex but repetitive work.

I guess I hadn't considered that before. So just to feed you back what I heard, the easiest step for them was access to internal information. Oh, where do I find this? Or an HR policy or whatever. Right. And then you're saying that because it had access to that knowledge, the next step was that it wasn't just request retrieval environment. It now, hey, I've got all this information. Let me start applying intelligence to what I've got in my database to identify. Hey, maybe this. Oh, we shouldn't be doing this. All of that stuff is that. I guess I had never really thought about.

Yeah.

That ascension.

It start becoming a real. Yeah. Human agent. I mean it's not. It's not replacing the humans, but it's like real human augmented intelligence and it's really augmenting the work. What we are doing on a daily basis. Yes.

Huh. Interesting. So what are some of the. I guess the standard stuff. If there's a business that's out there that is, in your opinion at least I know you have the stronger technical background, but for somebody that was that had never considered that they had the. Either the environment to deploy AI or the budget to deploy it, or the talent to deploy it. And now that that's changed, what would you suggest to a business owner who is saying, where do I start?

I would say start small, but big, but think big. So you always should consider some like larger or bigger vision for your AI, you know, strategy. But I would never recommend to start so big because you should start with some smaller use cases. Adopting AI in a very simple, you know, workflows or procedures and as well making sure your talent pool and the company and the people are ready for this AI adoption. Yes. Yeah. So you never should forget about the people who are inside the company and because these people would be adopting this AI. Yes. And the goal is here for them to not be afraid of the technology. Yeah. But rather being open for adopting this technology.

How do you guys address that? That particular part about the change management, about the employee concerns.

Yeah. I would repeat again, AI is not here to replace us. AI is to support how we make decisions, how we make things faster, better with higher quality, so that people can concentrate on a more complex decision making and can start just boosting our performance. Yeah, so that's to boost our performance each day. Yeah.

So for the audience listening to you and I, this is, you know, an exotic application, like oh my gosh, it's capable of doing all these things to Edwin, you know, what's really going on. You were there before this was launched. You understand it at a way deeper level. So if you're not scared, we shouldn't be scared.

Right.

Like as far as it replacing you. And I think that, you know, I, I'll, I'll tell you, I, I hear that a lot. AI is to, to augment and it's not here to take our jobs and that sort of thing. My, my counter to that is beware of someone who knows how to use the tools. That's the more likely replacement for you. Right, but I guess I just thought it was lip service from the media and things like that. But coming from somebody who has a deep understanding of what the true capabilities are, been there, you know, for a while. If you're saying that, then I, I think there's a lot more credibility to that statement because it would make sense obviously for anybody who's an AI enthusiast to say, oh, don't worry, it's not going to take your job, even if they had some concerns or they knew that, that it would, it wouldn't behoove the perpetuation of AI in the business and that sort of thing. But to hear you actually say that's not how you guys see it at all, I think gives that statement again, more credibility. So, aha moment.

A good, good, good point and good, you know, like what you said, Chris, like we should be more scared of the people who are capable to use AI, you know, in their work at scale, rather than just, you know, be afraid of AI. Yes. So people who are capable of using it and are, you know, capable adopting AI for their day to day work, I really, you know, that the future.

If you're enjoying this episode and want.

To learn more about how to start.

Using AI at work, we've made it easy for you. For just $1, you can have full access to the Chief AI Officer community, which will give you additional training, custom software, daily Training, calls on AI tools, using AI automations, getting more from your ChatGPT sessions and the business of being an AI consultant. Simply go to chiefai officer.com forward/insiders to accelerate your AI journey. Now back to the episode.

Yeah, they're gonna get more done, they're gonna get it done faster, and they're gonna get it done at higher quality. Yeah. So I guess at a minimum that would be the recommendation for people is learn this. Not just, not, not because it's exotic and it's shiny object or anything like that, or because you can do the fun stuff, make songs, make pictures, whatever, that's great. But really consider this application, as the title of this podcast suggests, on how to use it at work. Right. So do you guys, when you go into a company, is there AI strategy? When I say strategy, I'm talking about the context of, okay, the business has a strategy and someone has evaluated where does AI disrupt it, where does AI enhance it. By the time you guys get involved in companies, has that conversation already happened internally or do you participate in that conversation?

I think both.

Because you're dealing with, you're dealing with big companies, right? Like you're.

Yeah, I mean, in some cases we are engaged into this conversations from very beginning. I mean, because some of the, like, companies would like to hear this outside advisory, you know, like to, to hear this the company outside the organization and to make sure they are not making the mistakes that, you know, other companies made, for example, and we can advise on how to start in a better, faster and smoother way. In some cases, this strategy already, you know, created, designed and also influenced by people inside the company because a lot of also big companies employ very talented people who, you know, are experienced in AI and can also guide that these companies, you know, inside. In both cases, we take and participate in some conversations and we advise, you know, how, how to make less mistakes while adopting AI.

So just for context, the, the average business that you deal with, what, where are they in the strata of enterprise? Down to small to medium business.

I think 80 or 70% of our clients are like enterprises.

Yep. Okay, nice. So obviously they have bigger budget, access to talent like you guys like Deloitte, like other, you know, big consultancies that the firms below them can't afford. And just there is a limited amount of talent on the marketplace anyway, so they just can't get access to it.

Right.

So what advice would you have for those companies? Because same thing, they, they have some talent in their business for sure, but they don't have the big budgets to access the recognized thought leaders necessarily.

I would repeat, think big, start small. You can really think of very small and I would say repetitive Use cases and workflows that can be automated very quickly without, you know, those big budgets, long timelines, et cetera. You really need just to understand your, your, your business and challenges and the problems and the pain points inside the company and just try to map how AI can influence each of them. Yes. Because it's not really about creating custom solutions today. Yeah. It's really in 90% use cases, it's really understanding what is our pain point and how we can solve that with the available technology.

That's great advice. So you mentioned computer vision earlier and when I was checking out what Adepto does. That's an area of operation for you guys as well, correct?

Yeah.

How does that intersect with what your client? Like? Is that a completely different client than the enterprise side or is it usually because you mentioned aerospace, you mentioned aviation. That makes a lot of sense. But how about some of these other.

Industries, the computer vision, it's a vertical of the technology. I mean it's like some, it's. It's the part of AI. This is where we just, you know, not analyze the text or read text, we analyze images or we teach computers to see as humans see. Yeah. And there is also bunch of different use cases, I would say, in such industries like for example, manufacturing, where computer vision is making a real impact when it comes to the production quality assessment, when it comes to the logistics part, when it comes to the adopting also smart robots into manufacturing processes or some of the workflows. So there is a really like different bunch of different applications of computer visions that can be adopted also on your operations. Yeah.

Yeah, you're operating. You're definitely applying artificial intelligence in a much broader spectrum than I am. Fantastic. So this is making a lot of sense to me because Mo, like when I go into a business, I'm looking again to impact SG&A sales, general administrative opex, things like that. Not necessarily such creative application of what you guys are doing. So it makes perfect sense that of course computer AI could power computer vision that could identify the smallest defect in a chip or something like that. When I get into that scenario, I'll.

Give you a call. Yeah, yeah.

So because it's, it may come up. So I'm interested in how you guys are interacting with the clients beyond the technical side of things and in particular the. Or do they know what they need when they reach out to you, or are they saying, hey, come in here and tell us where it can work?

Both scenarios. Yeah, both scenarios. Because it really depends on what is their maturity level when it comes to the Data in AI. And we are not talking here about the whole company usually.

Right.

It's rather how this particular department or business unit which is reaching out to us is, you know, ready for the AI or you know, its usage. So, and we basically cover both scenarios. In first we, we came in and we look at the data, we look at the process and we say okay, this and this and this can be automated with that and that technology. In, in second case we came as, we come as a like technical implementation experts and we say, okay, this problem can be solved with this technology.

Yeah, very cool. I'd be fascinated to sit in on some of those meetings because again, the new user like to them it's typing or talking to a device and getting thought deliverables, not necessarily the physical nature of the things that you guys are working with like computer vision. It's pretty fantastic. AGI is one of those topics that certainly the stated mission of a number of the companies out there to pursue and achieve AGI. What do you think the timeline is before we see something to the point? Because, and I'll tell you from a business context, Edwin, I'll tell you, there's been a lot of chatter about agents from the, from the release of like, I don't know, early 2023, but nowhere near the capability. This autonomous agency in particular. Right. 2024, there's companies out there that say that they, they create agents and they have agentic activity, but they're not autonomous agents.

Right.

And 2025, from, from the business perspective and the conversations I hear, it seems that a. The word agent has entered the vocabulary a lot more, but also an expectation. People getting into this are expecting that autonomous agency, which does not exist as far as maybe in some, a few rare cases, but not, you know, commercially available mainstream to the average user. What do you think is a realistic kind of scenario related to timeline on capabilities of that type?

Yeah, I mean we expect 2025 to be the year where we will see this real disruption and maybe a real change from what we saw in 2022. Yeah, so 2025, I mean we real, we expect to see a really, really differently behaving model capable of more, you know, activities. And then when we saw starting 2022. Yeah, yeah, uh, we still do not expect that to be a fully autonomous AI. Yes, we still expect to be that limited and we still believe that will be. Need some adoption. What I mean by saying adoption, that means that the solutions that we will see we will be capable of doing a lot, but with the proper instructions Proper, you know, coding and with proper properly feeded data. Yes. But I don't see yet to be that, you know, fully autonomous general artificial intelligence.

Okay, this is, this was a helpful question because now, you know, we're at the level of artificial neuro intelligence. I can ask it something, you can do a very specific output. So this would be for the way for people to look at it might be that the term agent and its application in 2025 is just a more robust application of narrow intelligence. Yeah. Okay. So folks, if you are expecting agents to be that situation where you can just type something in and it takes it full stream down the, you know, runs the ads, writes like not there yet. However, Edwin, the architecture of that would probably be a series of agents that were chained together or in a flow.

Yeah, okay. Yeah, yeah. I mean you should think of agentic approach of something like understanding your query in a way that it can divide that into smaller tasks. It can run those tasks, it can make a decision while running these tasks and getting back to you with the sum result. Yeah, okay. That's really what we have today. But it's, it's, it's, it's really needs human control yet. Yes, yeah. But what we expect it's still to. We will have to build these chains of, you know, agents and as I said, we need to feed them with proper, for example, business data to make good business decisions. Yes. Or decisions that are in the chain of actions. Yeah. It can be easier to build agent that can shop for you because it sees your patterns. Pretty obvious task when it comes to the consumer action. But it's always will be more difficult to make sure AI can make proper and very, you know, good business decision. Yeah. As soon as doesn't have this big picture and content.

Yeah. Oh, interesting. Because different than the conversation that we had at the beginning which was create the bot with access to the company's information. Oh, wait a minute, I've got this pool of information. How can I act upon it intelligently? The difference is that we're now going to be in these smaller discrete environments where it won't have all of that external context within which to consider its next action or its step. Is that accurate?

Yes, yes. All the decisions that agents will be doing should be based on some context and data. And when it comes to the company data or its decision making, it's really huge context. Yeah, yeah, yeah. Of data. It's a bunch of conversations. Yeah. A bunch of meetings, et cetera. Yes. So I mean we moving small steps to this. Yeah, yeah. And we are moving to the. I think to the place where AI will really take this whole huge. You know, I can call it knowledge graph maybe. Yes. So like information structured in a way that machine can understand properly afterwards take some of the decisions. But I think it still will take some time. Yeah. To, to. To be in this place.

So again, not being a technical expert, let me, let me see if I get this architecture right or tell me if this tracks with logic. Based on what you've said, it seems like the best case scenario for a company would be that they would have essentially their own LLM with agents operating, you know, powered by whatever model. But, but this, the data set would be weighted towards their internal information. So you could still benefit from the agents. But it's not drawing from the 34 billion parameters or whatever in Llama. It's drawing could be but it's at least more specifically or initially considering the company's data.

Yes.

Yeah that's pretty sexy.

Yeah. Yeah. And the whole company's data is structured for example in a. Some again knowledge graph and its knowledge graph represents like company's brains. Yeah. And with some neurons connections relations school machine interprets that very precisely and it's really knows where and how to make those proper decisions.

So man, I do these podcasts honestly I do them for the audience more than I do myself because I feel very comfortable in the, the narrow application that I do generative AI and business operations and stuff like that. Right. This conversation has been outside of my normal dialogue because it's, it's forced me to. To introduce technical considerations that I just don't normally have a need to. Right. And not. Not that's. I'm not wired that way to be interested in the, the technical nature of things. However, I can tell you that this conversation has led me to have a, a couple of AHAs for sure related to the. This concept, this paradigm of agents. AI agents. Right. Like the clarity of the fact that autonomy is not expected or even necessary or inaccurate in the. The. The term agent that was a big one for me because I was. And then this understanding of. Yeah there's plenty the relevance AI and crew A. There's fantastic tools out there for the average user to be able to do some stuff right. And. But I hadn't considered. Oh my gosh, if I isolate or weight the data towards my own company's data, the absence of. There's no longer the absence of this context specific intelligence which I don't get. If I go use an off the shelf. Right. If I go pay 39 or $99 a month for some sort of a. Yes, I can get agency activity, but is like baby steps compared to what you're able to do when you're specifically targeting or waiting. Again, the data training on.

Wow.

That was. To you, that may be basic stuff, but to me, that was actually a big aha. I understand it a lot better now.

Yeah. Yeah. I mean, AI is exciting field. Yeah. And it's really evolving a lot in those last years and we hope to see that will be adopted more and more. Yeah.

So do you in particular, in 2025, do you expect to be. I don't know what. What would be. Do you expect to be doing a lot of you and your team in 2025 related to AI?

Yeah, I think agentic approach to solving some real business problems would be the main topic. Maybe not from the beginning of 2025, but I would rather see as AI in 2025. Not just, you know, like input output solution, like giving a prompt and getting the output, but rather making some decision or action in the background.

Nice. This has been fantastic. Hey, so as you've seen, I had an aha moment, which I love. We call it when people start thinking in AI, right. They. They make that, oh, wow, can I do this? That happened on this call. So thank you very much for that. Edwin, for me personally, how can. Do you produce a lot of. Or does your company produce a lot of research or thought leadership or anything? Because I would like to follow the progress on this with your company.

Yes, we are. We are producing regular newsletters and like white papers on our findings. We also doing the. Our own research on how companies are using AI in their businesses. Not always related only to our client base, but also outside of, you know, our. Our client base, but rather looking on the. On the market in general to give to the market this general overview. Yeah, and we really invest a lot in R and D internally as we have our own R&D department where we build and test. Yes, a lot. The technology which is available on the market, basically to understand and get this feeling what is possible with this technology.

Is that. Is that what we can find? Start plugging into that research would be at Adepto, your company's website. Okay. Wow. Well, awesome, man. Thank you so much for dialing in and taking the time out of your day. I imagine you're probably pretty busy and the time zone for sure. Again, this has been. I've got a lot to think about. Edwin, any closing remarks for the audience?

Yeah, it's exciting. Really. Yeah. And we live in really exciting world today. And I hope you know that our conversation also inspires your listeners. Nice. How, how, how to use and how to think, you know, about AI. And maybe, you know, they can also start exploring. Yeah. What is the possible and how they can just plug in that in their daily lives.

Yeah.

Yeah. Think, think, think, think big, start small. And that would be my, you know, final remark here, man.

That was great advice. And again, I had a couple of perspective shifts because this conversation is out. Like I don't have enough. I've got a great knowledge base on the operational and business side, the tactical, like the tactile side of business, not so much on the, the technical. And this has been a gift. Thanks so much, buddy.

Thank you, Chris.

Talk soon.

Thanks for tuning in to Using AI at Work.

Work.

Don't forget to subscribe for more conversations about how to use AI at work. And a special thank you to our sponsor, Chief AI Officer for empowering businesses with AI education and training. Visit their website for a free AI readiness assessment and AI strategy guide to help you get started using AI at work. That's www.chiefai officer.com. so thanks to our producer Evan Desalnier for making this this episode possible. Follow us on Twitter at the handle usingaiatwork and visit www.usingaiatwork.com for free resources to help you harness AI in your.