#234 Navigating the AI Terrain in SAS Businesses with Max Gutsche of UniFire.AI

0:00:02 - Mehmet
Hello and welcome back to a new episode of the CTO show with Mehmet. Today I'm very pleased to have my friend Max joining, from Germany. Max, thank you very much for being on the show. The way I love to do it is keep it to my guests to introduce themselves. So the floor is yours. 

0:00:19 - Max
Pleasure to be here, super excited. You know, cto show sounds like right up my alley. Yeah, absolutely so. I'm Max. I'm the co-founder of UniFire. I'm the product management, product business type of co-founder. So I do, you know, ui design, analytics and so on. And yeah, recently I did a bunch of things in tokenomics or crypto and came up with a startup idea, because I create a lot of content and I always, you know, like, have the issue of like, not posting enough content, not repurposing the super valuable content I create. So that's how we came up with, you know like, the idea for UniFire was just like a I'm trying to find a category, but you could say it's a repurposing platform or you need content scaling platform, and that's the main thing I'm working on and, yeah, trying to be, like you know, juggle linked in creator with startup founder at the moment. 

0:01:20 - Mehmet
Yeah, great, out of curiosity, Max, like what brought you to this field, I mean to the technology field. 

0:01:27 - Max
So I don't know, like I was always, I always had ideas, you know like. So I think my first tech business idea I had with I had when I was like 19, I was just walking home and I was like I really have this problem, like for me it was like I studied in different cities and I wanted to meet new people and I was like this is really difficult, I don't want to go to meet ups always. 

That's how my first idea kind of came up and then I was like, okay, but I don't have any skills in that space. So I was like, okay, when to when to you to me, thought like, okay, what's the easiest way, I'm visual, let's do some UI design courses, started reading some books, you know, and then I tried to study computer science, but that never really worked. So I studied politics and sociology. But you know, you know, use that free time, you know, by educating myself on analytics, on growth metrics and UI design. And then I mean my co-founder, we build a bunch of things, you know, fail the bunch of things. We did build it and they will come. You know the classic mistakes and build. We try to build products that are way too big for us. We had to feature creep, you know. We did every mistake in the book. 

And now it seems that for the first time, it's going to work. 

0:02:46 - Mehmet
Yeah that's great and this is, you know, the question that I ask every founder, and I ask founders even from non-technical and technical founders. So there must be and you touch the kids to base on it. But there must be something that you felt you know it's a problem that you are facing and other facing it when you decided to start Unifier. So, like, what is the genesis? I would say that you know. You said you know what I have to build this. Like you know, this is the problem I want to solve. 

0:03:23 - Max
Yeah, so yeah, I really had that issue that I was sitting on this, on these recordings and all of this content. And you know, the content was with there was with C's involved, hedge funds, really smart consultants, so right, like beautiful dredged script of like extremely smart people with really deep knowledge and insight, and I was like how, like this should be a blog post. I should create link and post out of it. But I don't know if you know that, but I used to be a podcast guy, so I do my podcast, I post about it once, maybe I post about it twice, and then I take the time to create like a really cool new episode again. It makes no sense like I have these assets lying around and they're beautiful, that deep, you know, like they we could do so much more with them. And then we started playing around with GPC and I was like, okay, gotta do this. You know I was like this needs to be there. Like I have an idle asset lying around, I need to utilize it. 

That's that's how we basically came up with this whole thing and yeah, we're doing this for seven months now, from like very interesting shitty words into now a more and more elaborate product. 

0:04:33 - Mehmet
That's, that's great, great Max. And you know, like again it's the reason I asked the question again is to inspire fellow people who you know they want to start building something. Like you know you, you solved your own problem first and then you've seen like many other people have the same issue Now, but of course the road was not all pink and rosy right. So then you must have faced some, some challenges while launching. So what were the these, I mean, the biggest challenges, and how did you overcome them? 

0:05:12 - Max
I mean some really big challenges are simply around, like human perception of markets, AI products and so on, Because, like I was, like we want to scale existing content right and generated. Where I, all the startups and we see a pouring money into, I can generate something from scratch. You know, out of thin air, I can create tons of content. 

And I was like you know, I always assumed the unifier ideas and obvious one, but I don't see many people building in that space yet. I personally assume 15 people going to be on that. You know as soon as the stuff comes up. So you know telling people hey, no, no, no, we're not an AI writer. You know like that was already a challenge. And then obviously you know, coming from that, all the let's say, ai and LLM style problems, for example. I'm not, it sounds robotic. You know, what should we build first? A classic one is like this could be for everyone. So how do I find my first ICP? You know the ideal customer profile things like these are really hard. 

Okay, got a call, okay yeah. So, you know, the classic one is Now like try to explain and position your product so hard. Then obviously, all the technical things. One thing I realized, for example, is like user experience and your eye design interaction, design with generative eye products is a lot different to you know, doing like traditional SAS in a way. And yeah, there's like a lot of mistakes you can do. 

0:07:04 - Mehmet
Why is that, max? I'm curious to know this, why it's different than traditional. 

0:07:11 - Max
SAS. In my experience what I realized was like you give a lot of control away from your, from the user experience, to like a language model, so like if you have. 

It's easy, right. You have, like, an input level, you have a, I would say, language model in the middle and an output level right. So in all of these spaces you can make a mistake and happen. For example, our users load some, upload something that may be way too technical for its bad quality or whatever it's already ruined, and then the language model you don't like, you do the prompt engineering and you have to iterate. It's very hard to get like one like thing that works for everyone. And then in the output level you know it's really dependent on what came in. So in traditional SAS, like a Miro or a Canva, right, like, I can control every pixel, I can control how the features look, how they behave. And in here I'm like, okay, a lot of things that are dependent like lots of dimensions. I call them dimensions of user experience. They just changed and are dependent on external factors I have to manage. 

0:08:18 - Mehmet
That's interesting, but do you think, are we going to see and you said like we are not seeing much but do you predict that we're going to see more SAS providers integrating AI in the future? 

0:08:34 - Max
I mean, I personally think there's a like an insane amount of places where you could, where you could put these, you know these, these language models inside right I mean notion did it? I don't know if it's like super valuable. Canva is a, I think, an obvious one. I'd love to just, you know, take my brand guidelines, my templates from Canva and fill them automatically in chat. It kind of works already. But even if we go deeper into tech tools, I found super cool as mergedev and other people trying out, for example, integrations for SAS founders are just a nightmare, right. Yeah, really like I should probably already thinking about the 150 most important integrations and they are now tools that take language models, they read the documentation, they look at the data schemas and they try to model both so that. 

I can get data from my app into another app and, you know, even even automatically code some of the code for the integration, saving huge amounts of time, right? So I think you know people think about the content and the information stuff. There's so much more we can do. 

0:09:44 - Mehmet
Now maybe I'll ask you a controversial question Now because of AI? I've read it somewhere and I think even I did when I was doing solar episodes. You know, like some people claim that because of AI, sas business might decline and the reason is if anyone can build something like using AI tools, so why do you need to rely on a provider to give you this service? What do you think about that? 

0:10:20 - Max
Um, yeah, I mean. I mean I think they have a valid point. One thing we realize already is, like you know, the large language model providers and JetGPT they will not build most of the functionality. You know, plugins are nice to have and they're like, they're really good for simple use cases, but it would, you know, destroy the whole purpose of a plug in it that they become really sophisticated software product. So obviously stuff like the content repurposing from Unifat is going to be integrated into Google Docs as well. 

But will you know, will Microsoft integrate short video content? Do the? You know, will they integrate scheduling and external social media analytics? I don't think so. It doesn't make sense for their business strategy. You know they would need to build an entire unit based on that. 

So, you know, I think, I think there's going to be a lot more consolidation, I would say, in SAS. It's like if you look at these 80 or 100 analytics tools that all silo the stuff and do the same thing, I think it's going to be very easy to just, you know, combine them into three or four Because, like, why, you know, like everyone can basically spin up these, these solutions. So, yeah, it's probably going to be harder, but the trend I'm seeing at the moment is the opposite, in people just launching hundreds of shitty tools. Like, 95% of these tools are absolute garbage. It's insane. So, yes, but I think it's going to, you know, it's going to be, with all hype cycles, going to be a massacre. 95% of these tools die off. And then there's going to be a lot of really cool SAS tools that that integrate and build with AI. 

0:12:07 - Mehmet
And, to your part, I think people are even big for these tools, right, yeah? 

So another thing when you build around AI and again, just, it's not like I want to hurt any brand, okay, let me not give the brand name so you know there are these tools that were built before ChalGPT went out and they do the copywriting right. So plenty of them, you know. We can name even some of them, like copyai, jasper and all these, and then, all of a sudden, you know what these guys have built became available without relying on them. So now, if I am a founder today and because, max, like you, have spent a lot of time on trying to get you know how to evaluate a good AI business idea how do I make sure that I can sustain as long as possible before one of the big names like open AI, maybe Google, maybe Microsoft, maybe Meta Facebook they come up and they embed this in their offering? So how we can plan this in a way that still we can have valid business? I mean, you know, on the long term, we need to be sustainable also as well. What do you think about that? 

0:13:37 - Max
So one thing I realized was like and this is not like something very elaborate, but so there's a lot of things or like dimensions LLMs are really good at, so for me there's like seven dimensions. In the beginning it starts with stuff like summarization, synthesis, so transforming information in any way. That's a really easy use case and these are like your meeting AI's auto AI. They're extric like ask my PDF, I think, on product hunt or whatever. There's 150 of them. 

It's easy to build and they are basically the LLM in a nicer UX. It's like as three buckets. We have a nicer UX, like from like 10 years ago. These are easy to copy. They're basically the LLM. It's pointless, can be replaced by plugins. 

But if we go now and go up these dimensions, so another would be categorization or extraction. So you have like a large dataset you find leads that fit to you, or crypto you want to invest in. So based on some patterns, then you come into like the more advanced abilities, for example, exploration. That's what the military is doing and the Chinese and the Americans. So exploring a military battlefield, supply lines, outcomes, you can. Even a friend of mine is exploring token economies. 

Or like autonomous agents that you really build on your unique insights and methodology. So, for example, if SAP builds autonomous agents on their resource planning data, their ideas, and obviously the more you move up, the more you move into personalization, representation, like you can be that resource planner or something then they become more unique, they become more complex and there will be really a representation of what you are uniquely good at and that's really hard to copy. Yeah, especially you know like, and once you can personalize that to the inputs of the user. I think you that's a business already, that's fairly unique and obviously you have all the other APIs of like transcription services, data analytics, machine learning. So only by taking different models. So maybe you can combine open AI and Claude, then you can combine transcription, image generation and you go for unique insights and methodologies you have. 

You're already in a space where it's like really hard to copy and you know, going from there, I don't think open AI will ever want to build something. You know, a language model or a process that's really focused on enterprise resource planning Doesn't make any sense for them if they want to be that compute platform where everybody builds on Right. So I think that's already a really nice way to think like hey, how can I make a really unique use case in the beginning instead of just ask another? Ask my PDF. 

0:16:55 - Mehmet
Mm, hmm. So this is where you know it's not about getting the idea, it's also about getting you know the validation and being unique. Right, max, like this is. This is the most important thing. Now you mentioned something interesting two things actually, so let me start by the first one. So you said you know, maybe one way to do it is to mix and blend things together and, like I know, like you, you you have the background phone on blockchain. Like, maybe it's a good idea to start leveraging, you know these LLM, you know large language models and API APIs to create a new use case with some other emerging technologies, like blockchain, maybe, or I don't know, maybe XR, vr, all these things. What do you think about that? 

0:17:53 - Max
I mean absolutely. So you know, like a friend of mine, a developer, he, for example, he built a use case for lawyers as though they can do trial, trial, mock, mock trial, mock trial is the word which are really expensive Usually. You set them up there, you know 10 lawyers and you know like a judge and it's like a role play simulation. And obviously, if you set, like in a high firm, 10 people with an hourly rate of 1000 bucks into a room for eight hours, you can imagine how expensive that is. So but he was like he used LLMs, put them into into like this VR application and, where you can, you can tell the language model. 

Basically, hey, this is the case, this is the mock trial. This is roughly how it goes. These are the participants you know, put on a VR set and go. It's very, you know, it's a very pure sea and it's like an MVP stage, but you can already see this is a use case that makes total sense for training, you know, for education. You know why not. So, yeah, that's, you know that's potential and especially if someone on from the legal tech side builds it, you know why not Right. 

0:19:08 - Mehmet
The other thing you mentioned. You mentioned about the autonomous. You know agents, right? So there are a lot of them. I mean a couple of months back, you know the first, like auto GPT, baby AGI, these, these frameworks came recently. I've seen the meta GPT also, as well as another one. How promising are these? I don't know if the framework is the right one, or is it like what, what should we call them? First and the second, like I've seen people who built full applications and full size platforms using these tools, like what do you see? What do you think about these frameworks, if we can call them frameworks? 

0:19:54 - Max
Yeah, I think I think if you build them and that's the big promise if you build them really on your unique knowledge you acquire over decades for example I don't know SAP does that for a manufacturing plant and resource planning like nobody has that much knowledge in the world and SAP Now give this, like make it an agent, give it to your consultants, give it to your junior consultants is going to be hugely powerful, right, like coding stuff seems to be. That's probably one of the I mean, it's not easy, it's crazy that it works Right. But you know, if, if we go into, like, the more advanced things, I think this is definitely a really cool, cool, cool, cool, cool, cool. I think this is definitely a really cool, cool way of you know doing things. And, yeah, I think they're super promising if you know, under the premise that you built them on something that's really unique. 

0:20:46 - Mehmet
Mm, hmm, um, like a little bit of a futuristic question what, what do you think you know the most promising development would be in this field of generative AI? And I don't like to ask now, you know, usually I used to ask people, you know, what do you think will happen after five years or like after 10 years? Now I think I should ask in the in the short term, I mean two years, because I know that many things will change. So what are you seeing from from your readings and observations? Yeah, I think I think meta. 

0:21:19 - Max
GPT is really cool, right. I think I think developer developers will have more, let's say, agent assistance. You know, and you know, especially, you know someone that is really good at documentation. You know, like co-pilot is already doing a lot of things. I assume that there's going to be something like meta GPT where you can just select a few agents and say, like I want this guy. You know, this is my documentation style, I'm training my documentation style. Maybe here's another one that you know is uniquely able to create data schemes for my integration or something. So I assume developers in the future will have, or like maybe maybe I have an analytics expert that I plug in and tells me Max, I think your data makes no sense, Right, like there might be an in as a. So I assume we're going to have some form of, like different set of these agents helping us in specific fields. 

I don't see them replacing us, like they can write code. But you know, like I tried a bunch of these UI tools, for example, and you know I want to do it in Figma. It's more precise, you know it's better for my vision, but I'm surely, like I, for example, need a support because I'm bad at typography and shadow and colors. Somebody could tell me look man, I think you're. You know your design system is just bad. So, like a design system agent that tells me, you know you want to be on brand with Figma and all the big ones. That's what you got to change. I think. 

I think that's going to be the next one or two years, surely? 

0:22:50 - Mehmet
Yeah, and I think to your point, max, the biggest I'm not sure if they how they're going to try to solve this out to your point and it's related to content. And when I say content, not only written content, I don't mean only text, I mean even, you know, pictures or images or videos, because at the end of the day, these large language models are trained on existing data, Right, so I cannot see, at least for the moment, unless there is a work in the background. And I discuss this with a couple of guests. Some of them they say no, it will be able to create, you know, to have its own creativity. 

You know, because, for example, now, if you look at the past two years, any SAS landing page, you know it goes by a little bit dark, you know it has this, you know violet color, but maybe this is the trend will change. So will AI be able to find something and how it can understand that all like humans will like it, because at the end of the day, I have to interact with that Like and this is, you know, the question is content creation, like. How is that changing? Because you know, now we are, we are getting, you know, the output from things that we provided input to. So what will happen in the? 

0:24:06 - Max
future yeah as a, I'm a strong proponent of, you know, stuff like what we do in now. The value of our, of this conversation, for example, will be exponentially higher in the future Because, like everybody can create a shitter of content. You know automated YouTube channels. They're quite good, you know they even like they work through proper news and stuff. But really, what AI is going to be, still like its limitations going to be at creating unique insights, new knowledge new frameworks. 

You know, like I could ask JetGPT hey, a value, tell me the dimensions of abilities and language models. You know, build me a framework out of it. I'm sure it's got something's good kind of gonna come out of it, but you need unique knowledge to even prompt that and ask it. So I think you know the 5% of the creators on top that are really like knowledge creators that create insights, that read a lot. They can use this just to scale to the moon. So I don't know that AI flood of content is gonna happen, but I don't think a lot of it's going to be 90% shit, so most of the stuff is just going to die. 

Also, kind of like the vision we are trying to do is just like we take someone like you and say, like you're a technical guy, you know you create amazing content. We just want to scale your insights, the way you speak, based on what you create everywhere on our platform. So that's really that's really the focus and I think that's going to be amazing with the eyes of you know, and I think it's also the reason why these jespers and stuff struggle because, like I tried to do it on my tokenomics content or gen AI, but it's just not good enough. It's like it's too robotic. It doesn't understand me right. It's good if you want to do a e-commerce landing page for tea or something, perfect. 

0:26:05 - Mehmet
Yeah, yeah, definitely, and the human factor will. In my opinion, max will be always there, because unless, again, as you mentioned, to your point, like if they change the way. How you know, there was a very famous thing in electrical engineering I still remember it, I struggled to understand when I was at college which is called feedback loop. Right, so if they, if they managed to get the feedback loop and again, I was talking to a guest the other day using neural networks and all this technology, they might do some breakthroughs over there. But still, without that, to your point, like, I tried a bunch of tools that generate videos, that they generate and logically they do perfect, but they are still robotic, because you know, you can feel, you know there's no creativity over there, and I agree with you and this is why you know someone. 

Someone jumped at me the other day and said you know what? Maybe? You know you don't have to do the podcast because there will be a you know some other AI agent. You just make it looks like you and he would be talking to people and even maybe talking to other agents as well. So I said, yeah, but you know it would not be authentic, it would be very boring and you know, I'm not sure these faceless YouTube videos. I'm not sure how they are able to make that impressions now, but I believe at some stage, max, you know, this way it's okay, something that will fade, because if you just repeating an information which is already existing in other videos, okay, okay, let's watch this cool thing. But again, I'm not sure. I think it's my own opinion on this matter. So, nevertheless, so, yeah, yeah, max, what other cool things are you doing? You know, like, what other cool startup things are you doing? 

0:28:01 - Max
I mean, like you know, I'm that creative guy so I always have too many ideas. One one thing I would love to do is, for example, you know, the B2B lead gen is an absolute pain. And yes, because, like, the way I target people is so specific that none of these databases can really do it. So, you know, I could imagine, for example, connecting to a link and API language model that, for example, scans the profile and tells me okay, you know, my customers could be people that do webinars, live events, you know, conferences, podcasts as well, so, and they have like maybe 6000 followers, they're engaged, blah, blah, blah. That's, for example, a product where I could imagine that would be that that's something, that something we actually might build internally, because it's such a, you know, such a pain to use the other stuff, and especially for someone like where, now I have a very precise idea of who I want to target. 

And the other thing is personalized learning. I think you know I have to extremely useless university degrees. For one international technology management in England, I paid 15k for outdated knowledge management. I still learn porters by forces and all this rubbish. So I'm like I'm always self-learning, I'm like doing cohort-based courses and stuff. So I'm like I think that traditional education is obsolete, but one reason why I hated school was it was so boring. It was always the same okay, we learn Spanish with Clara goes to Madrid, blah, blah, blah. And now I know a startup where you can learn Spanish, for example, and it's really focused on tech or politics or sociology or something that interests me. 

So all the lessons are, for example, personalized, what I'm interested in, and that would be like such a huge gain for me. 

0:30:03 - Mehmet
So this is super exciting. 

Yeah, yeah, I agree with you and we discussed it also here on the show the current model is obsolete, is absolutely obsolete, and I don't know how. This is usually what happens when tech starts to enlighten people. I would say they will resist for a while, but after some time I'm not sure how much is this sometime people will figure out hey, hold on one second. No, I'm not buying what you're telling me, and I believe this is something happening very, very soon, not how soon, I don't know, but yeah, and next, where people can know more about you and your work. 

0:30:52 - Max
Yeah, I mean. So people can always go to unifierai. That's the startup. 

0:30:58 - Mehmet
But it's really fun. 

0:30:59 - Max
Yeah, you're going to put that into the things. Anyways, it's with fire, fire, the burning fire, and then obviously on LinkedIn. I also try to go on Twitter soon because LinkedIn is yeah. 

0:31:14 - Mehmet
Or we should call it X. 

0:31:18 - Max
Yeah, I think it's going to take a while until people call it X. 

0:31:22 - Mehmet
Yeah, I still see people that they say, no, we're going to keep calling it Twitter. Sorry, elon Musk, we're going to keep calling Twitter, not X, but anyway nevertheless. So, max, really you enlightened us on a lot of points, and especially for anything, you want to advise anyone who's on the verge to start something in this space, like any advice you can tell fellow entrepreneurs before we close, I mean for us it was very. 

0:31:53 - Max
My advice is going to be very contrarian, because everyone in the startup world now says build something and be peers and launch fast. And I think in generative AI there's so many startups just coming out and they're all bad and people they're going to be tired of this. So for us we have 0% turn, because we spent the first three months for four months, really analyzing the market, talking to a bunch of people, solving our own problems and really obsessing over the first features and functionality. It's still hard, but don't just dump something shitty on the market and then expect people to give feedback. Everyone says, yeah, build, launch fast and get feedback, but if you want to have one thing, then build it well, take your time and then launch it to the market. 

0:32:51 - Mehmet
I think that's yeah, I think, max, this is. It's like early validation, before they'll build and throw things in front of people. This is 100% I agree with you on this one, because now, as we were saying, it's very easy to put guys. Let me confess to you something. I've put three or four let's call them micro SaaS tools and even I shared them on LinkedIn and guess what? It took me 30 minutes to build this up, but you need to see if there is a market for this. So, as Max, you're 100% right, and even we advise people not to put a lot of things without validating the market. So, and important also, it's changing the redscape. So I agree with you 100% on this. Anything I missed, max, I should have asked you. 

0:33:58 - Max
I don't think so I think we're doing pretty good. 

0:34:01 - Mehmet
Okay, good, it's not a tricky question, just to make sure that I didn't miss anything that you might wanted to say. So this is why I asked this question at the end. Max, really I appreciate the time. Thank you very much for being with me here today, and this is to the audience. Guys like, thank you very much for tuning in, as usual. Keep the feedback coming. 

I love to read your notes that you sent to me by email or sometimes on LinkedIn. I'm like Max, I'm on Twitter, slash X, but not very active. So thank you for sharing the feedback and don't be shy to write even things that you think I should make better, like. I love compliments, of course, but I love also, like, constructive feedback. So also don't be shy, and if you're interested to be on the show as well, don't hesitate to reach out to me. I'll, you know, be more than happy to discuss this and we can make the arrangement, because my goal is as much as possible. If you have a cool idea, you're working on something cool, or you have a story to share with fellow technology enthusiasts and startup founders, or to be founders or entrepreneurs, please like, bring that in. This is what we are trying to do on the show. And again, thank you very much. We'll meet again very soon. Thank you, bye. 

Transcribed by https://podium.page