Bridging Creative Vision and Technical Execution in the AI-Driven Economy with Sebastian Chedal and Donna Mitchell

 On this episode of the Pivoting to Web3 Podcast, I sit down with Sebastian Chadel, CEO of Fountain City, to explore how small and medium-sized businesses can actually use AI to solve real-world problems.

🎙️ Discover the essential steps to prepare your company for AI, the common myths and fears holding leaders back, and powerful stories about capturing critical knowledge before it walks out the door. Sebastian shares how his childhood fascination with tech and creativity grew into a mission to help businesses future-proof through AI workflows, agent automation, and strategies for knowledge retention—especially in a retiring workforce.

Plus: Favorite tech tools, smart contract readiness, and why AI alone isn’t enough—unless you know how to implement it successfully.

Whether you’re AI-curious, Web3-wary, or simply looking for smarter ways to run your business, this episode is packed with actionable insights and inspiring real-world examples.

Don’t miss out—subscribe now for more expert conversations bridging the gap between innovation and business transformation!

3 Key Takeaways:

1. Start With Foundations, Not Tech:
Before launching into AI, ensure your processes and data systems are solid. AI’s true value comes after you’ve mapped your workflows and organized your knowledge—otherwise, you’re building on sand.

2. Prioritize Quick Wins for Buy-In:
Don’t overhaul everything at once! Identify low-hanging-fruit use cases (like automating follow-ups or capturing “tribal knowledge” from senior staff) that show value fast and build momentum inside the organization.

3. Testing and Compliance Matter:
It’s not enough to simply launch an AI system—regular testing for accuracy, compliance, and user experience is crucial. Sebastian’s new venture specializes in AI/agent testing to ensure solutions aren’t just fast and cheap, but truly better.

 Visit [mitchelluniversalnetwork.com](https://mitchelluniversalnetwork.com) for more updates. 

#Web3
#AIIntegration
#DigitalTransformation
#BusinessAutomation
#ArtificialIntelligence
#ProcessAutomation
#AIForBusiness
#TechInnovation
#DataDriven
#CreativityAndTech
#FountainCity
#AIWorkflow
#KnowledgeRetention
#SmartContracts
#Blockchain
#AIAgents
#FutureOfWork
#BusinessGrowth
#InnovationInBusiness
#PivotingToWeb3 

 About Sebastian Chedal:
Sebastian Chedal is the Founder and CEO of Fountain City, a company specializing in AI implementation, growth marketing, and digital transformation. He focuses on helping businesses transform through practical AI solutions and strategic marketing, emphasizing measurable ROI and genuine partnerships. His key topics include addressing the AI implementation gap, moving beyond basic AI experimentation to real automation, and evaluating AI opportunities based on ROI. 

Chapters

• 0:00 Fountain City: A Journey in Creativity
• 3:17 AI Journey: Full Circle Realization
• 6:58 AI Strategy and Opportunity Mapping
• 11:41 AI Integration: Key Components and Solutions
• 14:38 Streamlining Sales Follow-Up Process
• 18:31 AI-Powered Engineering Co-Pilot
• 23:10 Context-Specific Knowledge Essential
• 24:34 AI and Smart Contract Integration
• 28:51 "Choosing Python for AI Projects"
• 33:06 AI and Productivity Revolution
• 34:50 AI Empowering Leaner Small Businesses
• 38:17 AI Testing and Quality Assurance

Transcript

Donna Mitchell [00:00:00]:
Good morning, good afternoon, good evening. Welcome, welcome, welcome to pivoting the Web3 podcast. And Sebastian Shadow from Fountain City is the CEO. He has a really interesting background. He comes from that creative place, but he's in the technology space with myself. And I'm very intrigued on his past and what he's doing now with medium to small businesses and maybe some large ones too. So let's check in, see what he's up to. And Sebastian, say hello to your audience and ours.

Sebastian Chadel [00:00:29]:
Thank you so much, Donna. It's a pleasure being here.

Donna Mitchell [00:00:33]:
Well, I'm glad to have you here. I'd like to know, and I'm sure everyone listening would like to know a little bit about your past and how creativity really connected with technology. We were laughing about. It's a little ways back, but I'm kind of curious to know how did you come into this space? And this is because I have a colleague, she was an engineer that became an artist and then the artistry brought her into AI. And her AI and web 3 background is how I really came into the podcast world because I didn't know what web3 was in NFTs. So I love the stories in the background. Can you share a little bit with us?

Sebastian Chadel [00:01:13]:
Yes, of course. So. And thank you for that question. So I, I've been my whole life really interested in both tech, technology in general, futurism, things like Star Trek when I was a kid and so forth and all the ideals around that, but also very creative. I was designing board games on paper cutouts when I was a kid and I taught myself how to program when I was around nine, with books. I got that I was. Because I was really interested in that. That was back in the 80s and that interest in both the intersect between tech and creativity evolved over many years.

Sebastian Chadel [00:01:56]:
I started Fountain city back in 1998 and definitely for the first 10 years there was a lot of, I would say, experimentation or following my passions or interests across everything from music with using technology to make music, but also designing, building websites, which was pretty creative at the time. I got really into Flash Action Script actually, which had an unfortunate complete death when the iPhone came out. But I became actually. So I was so good at ActionScript that I was being headhunted by large enterprise companies for a period of time because I could, yeah, I did really cool stuff with Flash Action Script, but. And so, you know, when, when we, when we, I, you know, the Fountain City really became a bigger agency that kind of intersect between creativity and technology continued. And back when I was also 18, I was very interested already in artificial intelligence and I, I almost went to university to study it. My life ended up going a different direction, but there was some interest there already at that time. And so my interest in AI, if that's where we're, you know, heading to now in the present, has been always there.

Sebastian Chadel [00:03:17]:
And now it feels as to me as though everything has kind of come full circle with the AI specifically because it's something that I was really interested conceptually in the past and now pragmatically with the work that my company does, it has become front and center and also applicable and appliable to businesses. Whether it's the machine learning, predictive modeling part, which isn't currently the sexiest part of AI. Everyone's very focused on generative AI and of course now all the generative AI and workflows and AI agents and things like that. So everything's just. Over the years, everything has just been constantly coming together. So in the, my first part of life, I was very divergent, exploring passion in different directions. And then as I'm getting older, things keep coming together where all my interests in the past all make sense now as they all come back in, you know, the professional space where we're.

Donna Mitchell [00:04:16]:
So it sounds like you're coming into your purpose, you're coming into your purpose in life, possibly with everything from your backtrack and your past. It all makes sense now. I could talk about that too, but this is not going to be our conversation right here. Everything's kind of making sense now.

Sebastian Chadel [00:04:34]:
I would, I would say so. But you know, life, life does go in waves. You know, there's. I, I think 2016 is when Fountain City hit. It went from just so prior to that date, it was me and maybe two other people working in. So it was a very small business. But then in that point, that's when we grew in six months to I think around 12, 15 people really rapidly. And that was the first moment where everything really felt like it was making sense.

Sebastian Chadel [00:05:04]:
All my different interests in information architecture and coding and design and servers and everything just kind of came together and it made, made. It was an easy trend or easy ish transition to have a business that size because I already kind of knew all the different disciplines and found City is not the only business I've had or have run. It's, I guess over my. I've had five nonprofits businesses up until now. So I currently have Fountain City on one other new starting business that's just an AI testing, but it's very, very.

Donna Mitchell [00:05:38]:
New so we'll have to talk about that as well. But first, what I really like to know is with Fountain City, what I had read is that you will take a company small to medium and you really get them the implementation. But I'm curious to know in the beginning, when you first meet, what does that conversation look like for those that are listening and have businesses, who is your favorite type of customer or what they say today, your avatar and how do you bring them into the AI workflow ideation process? Explain what your process is.

Sebastian Chadel [00:06:15]:
Yeah, and that's, that's a great question. Thank you. So I would say the first thing that I do before any work starts is really just understanding where they are at the level of knowledge, experience. Exposure to AI is so divergent right now. Some businesses, management owners know very little about it. People in the team are super enthusiastic. Sometimes you get the opposite. Someone at the top is really eager and interested in AI and the team is resistant or it doesn't, you know, maybe sees it's threatening or there's all kinds of different topologies, let's say in the, in within the organization.

Sebastian Chadel [00:06:58]:
So once I kind of get a sense of that, then the next thing I try and understand is do they already have ideas, aspirations, goals related to AI and then separate to that, do they have problems in the organization that could be solved by AI? So what are the pain points? And the third category would be are there opportunities or things they would like to be able to do that they're maybe not able to do right now? Maybe it's a limitation of their current business size or a new new area or market or things they want to move into or maybe they have never been able to get enough of their own content going or you know, whatever it is is the opportunity that they're looking to achieve. And so from there, you know, I also want to know who are the main decision area decision people, sorry, in the business. And then typically it then goes into a whiteboarding session that I will do and that I offer. So that could be with one person or it could be. I did one a few weeks ago that was with 12 people. It's a very big company. And in the whiteboarding session I'm really just looking to tease out those three things. So goals that they may have top bottom line problems and or opportunities and then start mapping out everyone's different ideas, throwing in also of course additional ideas of opportunities or ideas they may not be thinking of and really start mapping it in a matrix.

Sebastian Chadel [00:08:22]:
And the, and then the. Oh yeah, the other Thing I try and also determine usually in that session is is the business looking for a major transformation or, or are they looking for proof of concept, proof of validity? If it's the later, then we want to start small, quick results, get some traction in there, prove that AI is actually moving the needle in either top or bottom line and then build upon that success to do further projects. So yeah, the matrix compares against the two factors or the primary ones you cannot. The third would be time. But the main two are, are the impact potential of that initiative and the complexity or effort that that is going to take. So certain things are harder to do, maybe they have a higher upside. But you know, we kind of map it all out, look at how long things are going to take and then the end result of that workshop or whiteboarding session is typically a roadmap, suggestion or proposal of what can, what should come in what order. They can choose to engage us for those things.

Sebastian Chadel [00:09:26]:
We hope they will, but it is a discreet deliverable at the end of that that they can take and internally decide on, you know, what they're going to do in what order for AI implementation.

Donna Mitchell [00:09:37]:
So with the implementation do they need to bring in another company or you'll go through that process with them?

Sebastian Chadel [00:09:44]:
It is so, you know, there are limits to what we can do. We are pretty agnostic and very broad in what we can accomplish. So I would say most of the time we, we can actually deliver on the different things that they need to have done. I will say that a lot of the time what I discover is that businesses need to make two jumps or two steps to arrive at AI. It's not just one step that needs to be taken.

Donna Mitchell [00:10:11]:
What are the steps? Before you go further, what are the steps?

Sebastian Chadel [00:10:15]:
So the first thing that needs to be there is process systems. And some businesses I talk sometimes to pretty large companies. I think the most recent one, they were, they're you know, a 40 million. They're not huge like mid size, but they make 40 million annual and they have no CRM. So client relationship management database, no real sales process. Just calls come in and they give it to whoever is available. Everything's done by phone. It's all relationship based and it's awesome.

Sebastian Chadel [00:10:46]:
It's amazing that they've built the business already to that size. But if you're going to be automating or putting in agents or anything like that, you need to have processes and systems defined. So you know there needs to be a place to store if we're trying to, you know, if that's the area that we are being brought in or that is the need to automate and to bring in an AI agent. You need to know what your processes and systems are and if you haven't defined them yet, that needs to be done before you can or as part of the process in order to bring in AI. So you need your process and system. That's step one. The business may already have that and maybe it's just a question of improving or defining certain parts that maybe were done by people and it's tribal knowledge, meaning it's knowledge in their heads, it hasn't been documented yet or transformed into a flowchart and then the next big step, or depends on big or small step depending on the context is the Data. So, so AI's fuel is the data.

Sebastian Chadel [00:11:41]:
That's the oil of the, so to speak, of the AI operations or brain. And then the processes and systems is the fuel for the, for the actual automations or the workflows that the AI is coming in to either help, augment or perform. So those are the two main parts and depending on the organization and their need, they may have big needs in one or both or you know, or maybe things are already in place and you can just bring in agents for a quick win across that whole, that, that two step pathway. But yeah, so in terms of what they need to do in order to bring in AI, if their real challenge is that they don't have processes and systems yet, then how are we going to solve that for some organizations, one right now that I'm working with, you know, it depends on the organization. For this organization, they can't or they can't yet put in a CRM or kind of a system because of things that are going to be changing in the organization in the future. So we have to, so then we have to build a system that is very plug and play and can be easily replaced or connected to a different CRM later. You know, so it kind of turns into that kind of a situation. But, but yes, in theory.

Sebastian Chadel [00:12:57]:
You know, I mean I could list off, you know, different tech stacks.

Donna Mitchell [00:13:00]:
No, no, I'm gonna ask, I'm gonna ask you one thing before to make it a little bit more specific. I don't mean to jump in there, but I know they're listening. So please give us, give us an idea specifically. What I want to try to do is paint a picture for some of the businesses and brands that are, that are listening because a lot of people, there's so much fear out there. Or yeah, CEO, the front, they Just don't know what to do. So and. Or don't see the need. They know they got to do something, but they don't know the need, the solution, or the pain point.

Donna Mitchell [00:13:32]:
They just know they have to do something. They need to keep up. So what I'd like to do at this point is. Could you give us an example of a project that you walked into, where they were, what you did, and what the outcome was?

Sebastian Chadel [00:13:44]:
Yes.

Donna Mitchell [00:13:44]:
Is that fair? Don't have to tell us the name, but the type of business retail is, the pharmaceutical or supply chain, what was it exactly? So everybody can kind of wrap their head around what you're talking about.

Sebastian Chadel [00:13:56]:
Totally. Yeah. So let me give an example of an industry manufacturing client. They are. They. They're. They're my client that I'm talking about here that doesn't have full CRM systems in place right now. Yeah.

Sebastian Chadel [00:14:11]:
And in the whiteboarding session, there was a lot of things that came up. So the roadmap might be more than a year to implement everything. But where we're starting. I'll tell you where we're starting and what's coming afterwards. So where we're starting is with these quick win areas. So they have sales engineers, so they're technical people doing sales because they have to do custom solutions for their industry clients. So it's B2B. And they.

Sebastian Chadel [00:14:38]:
They are very. They. They need to be focusing primarily on developing these quotes that are, you know, engineering heavy. And the pain point here was that they end up spending a lot of time following up on some sales to see if the person's still interested or they haven't heard anything in a few days. They also want to be able to follow up with people who haven't done any requests in the last 18 months. So a lot of supporting activities were kind of a struggle, but they wanted to keep things still authentic to that person, meaning that it's not some other name or other person that's contacting or communicating. So. So we flowcharted out the whole interaction puzzle of how, you know, what are the different cases that need to be accounted for.

Sebastian Chadel [00:15:28]:
You know, someone says no. If they say no to. Why is it they're saying no? Is it because it's too expensive? So we act. So then once we kind of map out all the different pathways, it's basically just a flowchart. It can get pretty complicated, but it's just a flowchart. From there, we look at how much can we get the AI to do on behalf of the person? And. And then there's other, you know, non AI parts in it too that are just automation or triggers or delegation steps in there. So we look at also, you know, what is the most intuitive native interface for the team.

Sebastian Chadel [00:16:04]:
They want to keep, they want to make it as easy as possible to adapt this in. So they use Slack internally. So we're leveraging Slack for messages from and to the AI to the engineer that might them, hey, Bob wrote you back some questions. You should give them a call. Right? So then, then we're. The AI is triggering the person to make a human to human connection to answer the questions that they have. But if Bob writes back an email and just says hey, I want to go forward with the order, the sales engineer doesn't even have to look at that email. No response needed, nothing has to be done.

Sebastian Chadel [00:16:39]:
The AI writes back, says oh that's awesome, let me connect you to, you know, Susan, whoever that person's name is. And so it connects them over to that person who then takes their payment and schedules the work to proceed. And then if they're not responsive, the AI writes them, you know, two days later says hey, you haven't, you know, we did you this order two days ago. We didn't heard anything from you yet. Do you have any questions? And then based on what their questions are, the AI can just answer some of them for the sales engineer as well. So we're kind of augmenting, right. The sales engineer in this case with you could say lower risk client direct engagement over email and then it brings in the sales engineer when a person to person communication is needed to answer questions or to modify the order request questions about that. Otherwise I can.

Sebastian Chadel [00:17:26]:
The more exciting thing is the next thing that's coming up.

Donna Mitchell [00:17:29]:
But come on and tell us the next thing.

Sebastian Chadel [00:17:31]:
Okay, so the next thing that we're planning on doing that we're already concepting out is the knowledge retention of their current five sales engineers. They are I think a lot of organizations I see especially in industry manufacturing, but it's also true in other industries. There's a lot of older generation, older people that are thinking of retiring or getting close to retirement. And in their case they have five sales engineers who are all very close to retirement and they don't really have a new generation yet to kind of take over the, the function of doing this engineering role. So they know they need to hire new people, but they're very worried about these people leaving because they have 20 years of knowledge of, you know, building these custom solutions. And so we're planning out the strategy on how to do data capture and retention for their sales engineers. And so the end goal of that is twofold. The first is a system that the new sales engineer can just talk to internally.

Sebastian Chadel [00:18:31]:
So their own little private think of a plane co pilot. So you get a, you know, an engineering co pilot that sits next to you, which is the AI. And they can ask questions, for example, you know, what is a good material substitution for, for steel in this particular situation? Or you know, whatever kinds of engineering questions they may have. And then the AI is trained on these five engineers and they can answer back specific questions related to their company knowledge or the, the knowledge that they have. And then the other objective of having this system is when a person comes in and says, hey, I need a, you know, so we're back to the sales engineer doing their job. I need a quote. For this particular configuration setup, the AI can read the AI that is trained on the sales engineers can look at the quote and it itself can design the solution based on the knowledge of all the engineers and then present that to the new, you know, let's call them a junior, but they may or may not be junior, of course. But the new hired engineer will then get that AI generated quote and the engineer can look at that.

Sebastian Chadel [00:19:38]:
And you know, the goal there to begin with is to get the quote to be 80% accurate or more. And then the engineer can then look at it and make tweaks and modifications to get it to, you know, to use their insight and knowledge to get it to be exactly what the client needs. And that should also dramatically speed up the amount of orders that can be handled by one engineer per day. Right, and they're spending more time thinking about the, you know, you focusing more on the, the person's higher ability functions of like, is this the right approach for the situation? Should we be using a different material? You know, so they're doing a more higher role function of engineering. And then the, you know, the more mundane or baseline work of the engineering is handled by the AI itself so that people can focus more on the exciting parts in this case of the engineering work. There are other projects we have for them because it's like a year long roadmap. But those are like the top two that I can think of right now or you know, that I want to mention.

Donna Mitchell [00:20:39]:
Yeah, well, thank you so much. So let me ask you a question. I want to make sure I understood this. I think I heard what I heard. So, so you have the AI, these AI agents, they come in and they're going to be More efficient and effective. But when you started out that conversation, it sounded like you had an AI agent that would capture the knowledge base of those that were retiring. So you would feed them that information and then you would be able to transition that information, of course, to your new hires or whoever else you're onboarding. But they're available with that knowledge base.

Donna Mitchell [00:21:18]:
Do you feel that is any different than the normal process of AI agents that's out there now, or it's just the fact that you're using it more specifically for the cognitive benefits and the knowledge, let's say, of the senior employees that have been there 20, 30 years. Is anything different in capturing that knowledge base versus a regular AI agent? That's what I want to know because I'm one of the old folks. Okay, so I'm curious about this.

Sebastian Chadel [00:21:54]:
Yeah, absolutely. Yeah. The, the, you know, from going back to that concept of data is the oil or data is. Data is really the most valuable asset that a business has. Before, it used to be the processes of the business, I would say now it's the processes and the data. I mean, you could argue the data was always very valuable, but the data is becoming even more valuable. It's like the, like shiny.

Donna Mitchell [00:22:19]:
What do you call, what are you calling the data? So everybody knows when we say data, what is data to you?

Sebastian Chadel [00:22:25]:
So data is. So it is your data. It does encompass your processes. So if you have processes that are defined in your business, that is a form of data. But in this case with the sales engineer or the engineer in general, it's going to be, you know, the data could entail things like spec, so it could be product specification, could of within the company, but it's, it's also solution information. You know, when do we use a particular product in this or material in this situation or in another situation? There's, I mean, companies have lots of different.

Donna Mitchell [00:23:01]:
So is it the who, what, when and how of the business?

Sebastian Chadel [00:23:05]:
Who, what, when, who's doing what, when.

Donna Mitchell [00:23:08]:
How it happens, you know, how in.

Sebastian Chadel [00:23:10]:
The old days, it's also, it's also the data, you know, it's, it's the, the text file documents that they may have, the specifications, the user manuals, the. If you have, you know, if you have a certain process that the engine that, I don't know, the technician needs to do every single time that. And when, when light number three goes red, then they have to do XYZ instead. You know, all those things are documented somewhere and they form a base of knowledge. If you just ask a general AI that isn't trained in your business. A question that's engineering related, you're just going to get a generic answer about engineering. You know, how to build a bridge in general engineering terms. Yeah, but if your business is building, let's say, bridges, that's completely different.

Sebastian Chadel [00:23:57]:
When I was talking about just now, you know, the client I was talking about. But, you know, you have very, you might even have very secret proprietary information in your business that you don't want to have public. But if you want to have an AI system within your business that is private, that knows that information that can assist you, then you're going to have to train that AI agent within your company on your proprietary secret information that you have, your secret sauce so that it knows the answers to the questions that you want to ask it within your business.

Donna Mitchell [00:24:34]:
Well, I think that sounds pretty exciting because at the end of the day, it sounds to me that you have really been able to dig in and utilize those agents as best possible in the, in the fields that you have been working with. Have you ran into anything with blockchain or smart contracts at that point? I guess smart contracts, that to me, for anyone listening, is the intersection in the process where you have AI and then you also have smart contracts operating. If this happens, we want to do this or that, and it happens automatically with different variables and functionalities in play. So let me ask you this. So with everything that you have seen and been exposed to, how do you determine what workflows are going to be available to them than others? Do you have specific tools? I think you were getting ready to talk about your tech stack earlier, and I may have interrupted with a question and I'm curious to know, what are your favorite tools? We talk about that here on this podcast that you look at or like using or you want to feel comfortable mentioning and their functionality.

Sebastian Chadel [00:25:56]:
Okay. And you did pass through some other questions along the way before the last one, but I'll answer your last one about that.

Donna Mitchell [00:26:03]:
I might have forgot. I'm in that generation.

Sebastian Chadel [00:26:07]:
You asked about blockchain smart contracts, then I think you asked about how to. How do I. Or how do we derive decisions around workflows. And then you asked about tech stack or favorite tech tools.

Donna Mitchell [00:26:24]:
Okay, you want to go for it?

Sebastian Chadel [00:26:28]:
I'll go backwards. I'll go backwards. You'll see what the conversation goes. Yeah. So for the tech stack answer, so I will say first of all that I am one person within our company or, you know, right now I think we're around 18 people or so. And so we have different AI engineers with their own specialties, their own knowledge. I'm more, I will say I'm a generalist as opposed to a specialist. I'd say my specialization is that I am a generalist, but I'm pretty quick at picking up things.

Sebastian Chadel [00:27:01]:
So my personal tool is not necessarily the one that we're implementing for clients because I, I like tools that let me prototype things quickly creatively, sometimes in a messy manner, but it's very satisfying because it gets things done really quick. So with that disclaimer, put up front the tools that I like personally, Nathan is one of my favorites. N8N and that one is really great because I can build workflows or Agentix systems or MCP servers really quickly in a block line coding kind of environment because you're just drawing boxes and drawing lines between them. It does allow you to add code within the nodes as well. You can Type Python or JavaScript in them. So I like that. It lets me also get my hands dirty in the code when I want to go into that level. So it's really great for quickly prototyping things.

Sebastian Chadel [00:28:01]:
And I would say it's very suitable for a smaller business. It doesn't really. And it's also very suitable in private environments or if you need HIPAA compliance or other kinds of tight security because it's open source and you can install it on your own private ecosystem and just lock it all down. So it's, it is quite good. But it does eventually it's some of the main limitations it has for scalability of your team. Not necessarily the code is that because you're, it's not, because it's not code, you can't really track it in version control Git. If I can throw out some terms like you could, you know, just some regular old code that you have in Python or LangChain or Django. So it's not as suitable for a development team.

Sebastian Chadel [00:28:51]:
So in the end, even though I love building a lot of things in it myself and also for some smaller business clients it's great once you get to mid size plus the team, you know, pats me on, on the head and said that's nice but we're going to use, you know, we're just going to code this whole logic framework in Python. I mean Python of course is, is language that is very suitable for AI. We're using LangChain. If we're building front end apps then a lot of it is also still react. You know these, these systems might be coded by AI or AI Assisted and for auto. And then I would say the three main ingredients of an AI solution, there can be a fourth, but it's going to be you need your logic control. So that could be Natan, Python, LangChain, the thing that's going to decide when should an AI do what and what is the workflow, what is the flowcharts that you're going through. Then you need some kind of a data store.

Sebastian Chadel [00:29:49]:
And that might be traditional data, but it might also be a vector data where you're storing the data in more of an AI friendly multidimensional cloud. I don't know how familiar audience is with vector databases. And then you need the actual AI model and that could be a private open source model that is not on the Internet so you keep all the data secure. Or you could be using one of the big players out there like anthropic or OpenAI or Google's Gemini or something like that. So those are the kind of like the three main core elements that you'll see in an AI solution performing different things and then different connections to different other systems. If you're going to have a fourth, you know, maybe have your own MCP servers built into that too. It depends on what you're trying to do. If you're trying to do things that are very predetermined and deterministic, then you want more of a workflow approach.

Sebastian Chadel [00:30:43]:
If you want to do things that are more conversational with your AI solution, that's when you're looking more to doing a human to agent system that most likely will start pulling in MCP services. But you know, MCP is less, it's less deterministic. So you'd probably want to stay away from that in the, in the more workflow approach where you're trying to get very determined outcomes. Like if you're trying to get the AI to determine the sentiment of a customer's response and then dispatch the right action based on the sentiment it detects. So sentiment could be I want to buy, I don't want to buy, I have questions. You know, those would be a sentiment check, for example.

Donna Mitchell [00:31:23]:
So what makes you nervous with AI? What makes, what makes you, what worries you when you go to bed about AI? If anything.

Sebastian Chadel [00:31:33]:
I mean, that is, there's, that is a very.

Donna Mitchell [00:31:37]:
I had to ask.

Sebastian Chadel [00:31:38]:
No, no, it's, it's an interesting question because there are so many different things within AI to think about. You know, sometimes I worry about my friends that over worry about AI. Sometimes. I mean pri, I would say last year I Worried that if we, if we, meaning our business didn't really change quickly, that there are certain industries that are in trouble, you know, and I think the traditional or the prior.

Donna Mitchell [00:32:07]:
Which ones do you think are really in trouble?

Sebastian Chadel [00:32:10]:
Well, so the one that we were in was the one that I think is in trouble, which is just the agency that you know. So for many years we were building websites, also helping businesses with technical digital transformation. But a lot of it was web applications or cross platform applications and websites. And I think that sector is, I mean not only has that sector become very saturated, there are lots and lots of agencies out there, but the, we were already seeing pressure in that industry from SaaS applications that are becoming more and more performant. You know, think Shopify for example for building your online store. But also we were seeing the democracy of tech skills democratization, I should say sorry, of technical aptitudes increasing globally. So more and more teams outside of the US are getting better. Really good.

Sebastian Chadel [00:33:06]:
And then you have, you know, currency exchange rates to, to deal with. That's. So there was already this feeling of a race to the bottom in that field. And then AI just kind of amplifies it even more because I mean we can, the things we can do now with our productivity output with since we've been bringing AI into our team is just, it's really big. I mean we, the thing, you know, it used to take maybe two weeks to do a prototype that was just a click through screen for a client to show them what an app can do. Now with a team that's a third of the size, you can make a real app instead of a clickable prototype in the same amount of time and budget in two weeks. So it's a third of the team. And you, and you could argue the, what you're delivering is way closer to a final product really quickly.

Sebastian Chadel [00:33:58]:
So. But I do think the, I do think what we're seeing and going to be seeing is also there's different trends I think they're going to pick up. One of them is the micronization of AI models and their integration into everything. So you know, right now the big focus is a lot on these large language models. But I think there's a future coming where you're going to see these really small micro models, so to speak, that are trained in just very narrow domains which keeps their file size a lot more manageable and smaller. And I think, you know, techniques will come around to make the file sizes smaller and smaller and or devices, you know, keep increasing their storage capacity. But you know, imagine that every single app on your phone, for example, has an AI model in it. And that AI model is really specific within that domain and it can be offline.

Sebastian Chadel [00:34:50]:
Right. You can just store it on your device. And that kind of approach that thought of like there's going to be all these micronization of things, I think business, there's going to be a, in a way, a bigger way, a big wave coming of smaller businesses that are, that are empowered by this new economy with AI. And so we're seeing the kind of this weird, I think, double wave where you've got large businesses that are trimming down and laying off a lot of people I'm seeing for various reasons. I mean the thing is the economy is always complex. Right. You can't just say it's AI, but there's, there's generally this, this layoff wave that's, that's hitting larger businesses and then smaller businesses that are in the older paradigm are struggling and so they're also, some of them are folding and being absorbed up. But there's going to be a, this other wave that's coming of these smaller businesses that are, you know, maybe it's a business of 15, 16 people, but it's doing what an old business could have done when they were 64 people or 32, you know, and so the, do you have leaner businesses that are able achieve more with the number of people they have? And because AI overall I feel that it empowers people and it also allows people to be doing more advanced or more thoughtful or more interesting work.

Sebastian Chadel [00:36:12]:
And. But it, the thing is, it's moving so fast and it takes the economy and people time to adjust. And so there's like this pain period in the middle of it as well. So. But I can keep going with things that keep me up at night because there's, there's a lot I could talk about. So there's. In different categories. Yeah.

Donna Mitchell [00:36:32]:
So we'll have to have you back. So before we close, is there anything you wanted to share that we didn't touch on?

Sebastian Chadel [00:36:40]:
Let's see. I mean, yeah, I think the, the new business that we're starting up is really.

Donna Mitchell [00:36:45]:
So what's the new business? Real quick.

Sebastian Chadel [00:36:47]:
Yeah, So I could just briefly mention that. So fantasy is really focused on what we talked about, which is helping businesses to move from idea into execution and implementation for AI solutions. This new business is focused just on one narrow thing which is testing AI agents and or chat systems to make sure that they're compliant, error free, doing what you expect them to do, but also that people have a great experience with them. Leave that experience thinking, wow, that was actually maybe even better than it would have been if I talked to a person or for an internal system, you know, making sure that it's correctly referencing the right knowledge and information. So we have a whole test suite that we help build and plus industry standards for these testings, testing suites that we set up. And, and then we can also do monitoring and of the actual systems to make sure that as logs come in and people, you know, if new issues come up, didn't expect people to be asking this or it's something new in the news. And now people are all asking these questions that the AI agents are, you know, that we flag, hey, there needs to be new knowledge or new alignment or maybe there's a new jailbreaking attack that comes out. It's kind of, it's another way of thinking of it.

Sebastian Chadel [00:38:01]:
It's kind of like a security test or security audit you might get for your website, except now it's for your AI system.

Donna Mitchell [00:38:07]:
That sounds very cool. So what happens if they don't make compliance? So there's compliance in that industry and you've got the AI? We flag it and they flag it. What happens happens.

Sebastian Chadel [00:38:17]:
So we flag it. They know that it's a problem and then presumably they have people in their company that can then fix the problem and then we retest and show that it's been fixed. Of course, if a company comes says thank you for testing all these problems, testing all these things, you found these problems, but hey, can you help us fix them? You know, we will, you know, we're available to also fix those issues. But the core of the business really is just to help people to test or for, for developers or other agencies that are building these systems, we can help come in as their support partner to ensure that the work that they're delivering and or for the client to ensure that work that's being delivered is passing tests and is effective. And a plus quality AI in general, if you think of AI you've got, or in any technology you've got, faster, better, cheaper, as the kind of the three holy grails that you're trying to hit in that new, that new revolution with AI, getting faster and cheaper is easy. I would say that's the easy too. But getting better is where it takes energy and effort and that's where you need to put in the actual elbow grease to make it actually be better than what you used to have.

Donna Mitchell [00:39:28]:
Right.

Sebastian Chadel [00:39:29]:
And that's where the testing really comes in because if you're not doing test driven development or test driven work to ensure that you're getting that A plus quality, quality, you're going to end up. I mean, there are so many AI systems that we test that just fail really badly and then people end up thinking, oh my God, AI is not ready. What is this? I was sold as hype. It's terrible. But if you're not testing the system with a high expectation of results and then not doing the work to get it to that level, and you're only putting in minimal effort, then you're not going to get the results and then you'll just be disappointed.

Donna Mitchell [00:40:05]:
So what's the name of the new company?

Sebastian Chadel [00:40:07]:
Doesn't have a name yet, unfortunately. We have a few titles.

Donna Mitchell [00:40:10]:
We're going to keep in touch. I'm really interested. I'm really interested about this one. I really am. How can people reach you?

Sebastian Chadel [00:40:16]:
So, website is Fountain City Tech and then I'm also on LinkedIn. My LinkedIn is just my first and last name together, Sebastian Shadal. And I mean, that's more than enough ways to reach me. I think either of those two would be great. And you can write us a message on our website if you're interested in the testing services, you can just contact us through there for now.

Donna Mitchell [00:40:38]:
Yeah, we'll find you. I'm going to stay in touch with you. You're not going to get too far away. I like that idea. So thank you so much for being with pivoting to Web3 podcast and we're shaping tomorrow together. Thank you, Sebastian Shedal.

Sebastian Chadel [00:40:52]:
Thank you, Donna. Been a pleasure.