Inside Treeo ai with Taymour El Kady How AI Powered Agents Are Revolutionizing Team Collab

Honestly, like I've been working with the these agents for four months like in an intense way, right? And just like you said, I'm not quite sure what they are. If I were to explain it, I think it's just guiding the LLM into a more focused direction. Usually LLM's hallucinate uh when they have too much on their plate, it's very easy for the AI to take another direction. And the way I see it is just with a few shots prompting and then with also telling the model what they are or what they're supposed to be doing, it really helps calling it an agent. But if if I'm going to strip all the technical part away, I think an agent is is just automation. You know, you're just automating a process. >> Hi, my name is Demetri Bonichi and I'm a content creator, agency owner, and AI

enthusiast. You're listening to the AI agents podcast brought to you by Jot Form and featuring our very own CEO and founder, Idkin Tank. This is the show where artificial intelligence meets innovation, productivity, and the tools shaping the future of work. Enjoy the show. Hello and welcome back to another episode of the AI Agents podcast. In today's episode, I have Tamour Elicadi from Trio. How you doing, Tamour? >> I'm good. How about you? >> I'm doing awesome. Um really appreciate you for making the time uh setting this up with us. We we really appreciate it. There's a lot of really cool AI tools out there and we found you and found what you were doing and just figured, you know, might as well reach out and and make it a make it a conversation. It's been um it's been very fun doing this podcast and we're

really excited to learn more about uh what you're doing at Trio. So, I guess just to start it off, tell us a little bit about you and your background and how you got to um where you're at now. >> Okay, so uh yeah, if we go uh way back, I I started my career in uh investment banking or ass management. Uh I kind of hated it, but uh it was a it was like a very interesting couple of years. Uh and then I went straight to startups in in Egypt at the time startups were uh up and coming and everyone is well not only in Egypt around the world everyone was raising crazy money uh when investors had deep pockets right so it seemed intriguing uh and I joined uh a trucking company actually it's a like a tech marketplace uh focused on trucking uh

let me let me tell you like trucking is at least for me like by far the most uh exhausting industry I've ever seen. Um but yeah, so I've I've I've played some different uh different roles from operational to product uh um also data. So I was like all over the place a bit. Uh and then uh I moved to another startup in Egypt also. Uh it's also a marketplace but focused on uh the hor sector or the supply chain for restaurants. um where I was leading the product team there. So throughout um my 8-year career in startup u for some reason the data team was always reporting to me. So I'm very familiar with how startups deal with data and this this was the the the initiation of of trio. Uh I was actually in a in a meeting in the leadership meeting in the

last startup I was in which was supplied uh and they're they're technically a a super technical uh company or group of people let's say and we were still negotiating and and talking about what churn uh is or what's the number right uh everyone had their own definition and then this was this was the turning point at least from a from an ideation uh point where I thought all the tools out there whether it's Metabase, PowerBI or all the BI tool, they're not really solving uh startups issues when it comes to data. They're usually solving for big companies that just need to visualize data, look at patterns, make decisions. Uh everything's more uh structured in a way. Uh so yeah, and then came uh Trio, which uh which Yeah. >> Yeah. No, I think that's a good that's a good place to get started. So, what

necessarily did uh you um what problem are you trying to trying to solve at TRIO and what is um you know what got you started with that specifically? Like what what um triggered you to think this needs to be a problem that gets solved? >> Yeah. So, a bit unconventional, but I would say I'm solving a few issues. Uh because I'm still getting started, right? you I would be trying to tell you I have one focus group and one problem to solve and I'm scanning. That's not the case. Uh but it all started with data inconsistency, right? That every every single team has uh has a different definition for each metric or KPI. Uh so there's always inconsistent data within startups. So this was the the first problem I was I was trying to solve. uh but then a few a few things uh a

few things came up and I wouldn't really introduce hundreds of features to solve different problems they're more or less under the same umbrella but there's also the fact that there's usually a bottleneck uh effect in startups with data teams that's that's that's generic not only startups but plenty of requests come from multiple teams to one data team and then the data team has to prioritize and then uh push push the requests according to uh whatever priority uh they have. So yeah, the second one if if I'm if I'm losing focus here, the second one is uh the bottleneck. Yeah, so data inconsistency and bottleneck. And then uh uh last but not least is uh there's so yeah I gave the the the churn example but uh I think I I just lost my thought here but but yeah just for for simplicity let's just say

these are the the main two focus and then there is a security issue where everyone exports all the sheets all the let's say I don't know the data and puts it on sheets, whether it's Google Sheets or Excel sheet, uh, and then plays around with it. Yeah. So, no matter how good the tool is, uh, the BI tool, people just find a way to work on sheets and then there's this crazy uh, amount of sheets and what's the source of truth and then it it goes back to uh, data inconsistency and and security. >> Now, I guess what are I mean, so first of all, how long have you guys been doing this? >> So, it's uh, it's been very fresh. It's been four months. >> Okay, that's totally cool. I I I saw that it was like about five months on your LinkedIn. So,

I was curious, how many people are you working with or you just bootstrapped yourself right now? >> Yeah. So, actually I started off with uh a partner, a CTO uh actually a CTO and uh um CDO or like there's no like someone with the data and AI and then another yeah handling the product. However, uh the the CTO uh has is is a has another company. He's a founder of another company and uh and then they just fundraised and it's a whole story. So, he didn't really uh continue and then um and then the AI person uh also got a raise in and sorry um uh what's it called? He got another role in the same company he's in which is very high high ups there. So, I'm all alone now. I'm a solo founder at the moment. Uh I have a team of uh

four interns. Uh one data scientist and then one uh developer. Uh there's also a part-time designer or project based designer. But yeah, so it's basically I'm bootstrapping up until now. I did a very quick uh or very small family and friends uh uh yeah and uh and yeah, that's my status quo at the moment. We're we're launching in hopefully hopefully fingers crossed two three weeks. Uh that's uh tops. We have like four design partners at the moment or like design partners just early customers that are willing to try the product for a month for free and then if it pays off then they're going to uh subscribe. Uh I hope by the time you release the you release the podcast trio would have would be launched. Um but yeah that's uh that's it. >> Okay. Very cool. Um, what is something that with this tool

you want people to be able to do? And like what what is the type of specific market I guess that you're targeting like size-wise of company and whatnot? >> Yeah, to be honest, there are a couple of go-to markets in in in the in the plan I want test uh at the beginning, but the the initial or like the four design partners, they're all let's say startups. They're all in the supply chain in a way uh in different industries but very operationally heavy. Uh and I'm I'm really tackling nontechnical people. Uh having a a a chatbot in the in the tool is is not the only thing handling or dealing with technical people. Uh I have for example a sheets interface where you can uh communicate internally uh internally with it. you can handle like there's a training mode where you can add your metrics,

KPIs, business logic, whatever. So you basically improve the AI knowledge layer in a way. So the tool is more tailored and agile uh to your own uh to your own company needs. Yeah. But just to circle back to your question, they're mainly operationally heavy startups. Uh but generally anyone that has a database is potentially a customer, right? >> Okay. Um, and I guess with this, you know, obviously there's a lot of data uh security kind of and things that are going on like you're in the EU right now. So I'd imagine with with this little uh this whole topic like what's the stance that you guys are taking on like data and um people's data. I'm guessing you're not selling you're not selling people's data, are you? >> No. No. All the data is is mostly encrypted. technically I I I would have access to

it but uh I think the the the issue here is which AI model are you using right uh and for example in the Middle East I for some reason they they prefer open source and not closed source I would have thought otherwise but open source is is u is more is is not under one company right so this is why they they prefer open source while in in in Europe they prefer they prefer closed source and then there's also this ranking I'm not 100% sure of this fact because I heard it from from someone but that someone works in uh the Danish government in in AI and apparently they have ranking for each country uh countries AI models right and then it's uh yeah and uh and yeah China and and and the states are somewhere and then Europe are somewhere so they're also very

promoting uh European models Right. Um so I think we don't have an issue there uh because we can play around with the models with AI models uh according to uh the needs right because the agents itself the architecture of the agents or how they communicate is the most important part and then you can just plug in whatever uh whatever agent uh makes sense right. Yeah, that's fair. And I guess what are your thoughts on, you know, you're talking about agents now. Um, what are your thoughts on the advent of, uh, agentic AI in the last couple months? I feel like this entire year has just been like off the ground. Um, it went from I don't even know what an AI agent is to it's the the biggest buzzword on the planet. >> Yeah. Honestly, like I've been working with these agents for [Music] four

months like in in an intense way, right? And just like you said, I'm I'm not quite sure what they are, but uh if I if I were to explain it, I think it's it's just guiding uh guiding the LLM into a more focused direction. Uh usually LLM's hallucinate uh when they have too much on their plate, uh it's very easy for the AI to take another direction. And the way I see it is just with a with a few shots prompting and then with also uh telling the model what what they are or what they're supposed to be doing uh really helps calling it an agent. But if if I'm going to strip all the technical part away, I think an you're just automating a process. Uh it could be it could even be an algorithm. It doesn't necessarily need to be AI. >> Okay.

So, I want you to parse that out a little bit because I think I probably understood what you said, but I'm curious if the audience maybe didn't. So, could you parse out a little bit more about the whole hallucination topic and how because my my perception is that it's basically and I use agents a lot. It's preventing the hallucinations because of a framework that it gives or tools that it adds to the agent. You know, usually um that prevents the hallucinations because it's guided within a specific task framework. is uh would you have anything else to speak to that or how how would you kind of explain that to somebody uh what you mean by uh it's preventing hallucination because you said if it has too much what does that mean? >> Yeah. So I think uh the the rag model has was was has

been recently again upcoming. I think notebook LM was the first at least commercialized uh rag model and and it's just a a knowledge layer. It's just a very fancy word for knowledge layer, right? And I I think this answers your question a bit and and the hallucination again when it's too much like we said it you want you want AI to focus, right? So what you do is you try to strip whatever it knows uh other than uh the natural language communication part and just give a a bigger weight uh to whatever you're feeding it at the moment. Uh and yeah this is how I I I would explain it. Uh however I think from a technical point this is not what happens but yeah uh I think it's a good explanation. >> Okay. Yeah, because um I don't know if you're familiar. I use

tools like relevance a lot. Uh relevance AI is a pretty big agent tool right now. Obviously, Jot Form AI agents um as try Jot Form AI agents using the link down below. Um, but I I found in the last six, seven months at this point as we've talked more and more about agents that they went from essentially just a custom GPTs, you know, where you kind of had guide rails about how it would act and then you started adding tools on top of it. What are your thoughts on like the new addition of tools and the access to the internet that agents have? Because for me I remember thinking last year the big gains that we'll see in AI in general will be when chatbt any LLM etc are going to have internet access in APIs which is just like commonplace now whereas four months

ago I would have been freaking out. So what are your thoughts on the tool additions to agents that have uh been added in the last couple months? So you mean just so I I understand your question, you mean like the web hooks are integrating with these agents too? Uh yeah, so web hooks I guess is one of the examples, but like for for context, um I would say API HTTP access in general connected to an LLM in the context of internet search and other and any tool, right? Like that's what I like about relevance a lot because basically you can connect it to anything that has like an HTTP uh API endpoint. Um, and that's essentially all these tools are whether they're easy to use or not, right, is one thing, whether you're technical or not. But relevance is the first example I've seen where

it gives you pretty easy access to, you know, connect to Perplexity, which searches the internet, or connect to a LinkedIn scraper, which uses the internet. That's what I mean by tools. >> Yeah. So, so yeah, just connecting APIs with tools and then just uh doing whatever you want basically, right? >> Mhm. Yeah. >> Yeah. Well, to to be honest, this is like a very uh let's say focused group of of agents, right? And it's it's all it's all coming from the from the from the same part. I I don't feel like there it's just everything is being rephrased uh or or said differently, but at the end of the day, this is a very uh niche agent, right? And my opinion this is going to be the direction going forward like LM are going to be extremely uh focused on specific tasks. Uh we're already

we're already seeing that. So I don't know like I haven't tried relevance to be honest. Uh but while you were explaining it I just I feel like it's it's a very similar uh thing to agents like it's not uh not not new to to be introduced you know. Well, I mean I would say that um my point was, you know, you see it now in the UIs of things like this, right? Like when you're doing um an 03 input on Chetchip or whatever, like it has the ability to essentially scrape the internet. Um I know it's fine-tuned for Google search or whatever is like the claim. Um and then it just does reasoning continuously until it gets the results. But my my my general question is like what are your thoughts on how that toolkit will expand and what that necessarily means for like agenta

capabilities in the next uh in the next year cuz it kind of came out of nowhere in my opinion. Sure, you could connect a bunch of APIs uh in a sequence in like make.com or something, but it's completely different than when you just make a individual chat into um a reasoning model that then comes out with, you know, the results that you're looking for cuz you'd have to set up like the meticulous step-by-step, you know, make.com or nadn workflow in order to get the result you want. But now you can type something out and it does something. So that's that's what I'm saying. I think that was the improvement. Where do you think more of these tool like additions and and whatnot are going to help um moving forward? >> Yeah, honestly, uh I I haven't really thought about it that much. Like yeah, just

uh AI in general, it's very hard to predict where where we're heading. Uh but yeah, just honestly like I it's sad to say that I don't have a lot to add here. Uh but um well yeah like I said it's just going to be very easy and very focused until we need to control it uh control it a bit right but right now I think everyone's trying new things and everyone is experimenting and we're just finding easier ways to to to do stuff but at the end of the day there will be a stop at least from a pricing point right there is a there is a a pricing bubble at the moment with uh with AI and then having that uh that easy to just again use all these tokens or all these agents to do something that's very simple that you can do

uh more manual. I think that's going to fade away as we as we go like as everyone knows like maybe chat GPT or Gemini even they're not making money over the millions of customers that are currently using. I myself I burn so many tokens on chat GPT. I'm 100% sure that the $20 I pay are not are not enough. >> That's fair. I think that that's a good comment. So what do you think is going to happen then? Can you like tap into that a little bit more? What do you think is going to happen with these uh tools then? >> Yeah. So I I just think so it's the the the major or larger companies they're gonna it's going to be monopolized some way because building LLM from from scratch is becoming impossible from a from a cost from a costing point. So there

will be the the the the unicorns of LLM. I don't think it's going to be easy to compete with that. Uh and then it's just layers after layers. So uh there there will be more tools uh to use these let's say uh LLMs and then it's going to become more uh more niche. So there will be a very focused LLM that's or or agent or tool at this point to I don't know like record this podcast for instance. Uh and then we're going to be just super focused. You're going to find a tool for anything. Uh in my opinion this will uh this will crash. this will stop. Uh because AI again is going to be super expensive and then people always think that AI will replace like head counter or labor or whatever it is. But this is in first world countries in a

way, but if you want to automate a warehouse in Egypt for instance, it's going to cost much more uh than uh hiring tens of people to to manage this warehouse. So yeah, just um I'm getting off topic a bit, but just to say that >> No, it's okay. >> Yeah, I think we're going to have a very super focused uh niche LLM or like solutions uh and then they're going to be super expensive at some point because the they're just going to charge much more and it wouldn't make sense financially. >> Um yeah, that's an interesting uh thought I hadn't heard before. I I have heard that they think I people think that they're going to get more niche and more specific. But what makes you specifically think that it's because so your your argument is essentially chat GPT etc are charging under what their

you know needs are for keeping capital you know they're burning right now and they're going to hit a point where they're going to need to start charging more. So, you're saying they're going to start charging higher prices for very specific niche uh agents once they get to the point where like the training data, I'm guessing, is enough to achieve that uh results that are good enough. Is that the argument? Probably. >> Yeah. I think there are two separate uh so one is a is an output or result of another. But the reason we're going to be very focused and and and very niche is because because of hallucination in my opinion. So, you just need you just want a tool that that does one specific thing. uh and as as consumers this is also something we like or something we get attracted to. So someone

sells you something this is what it does even though you can technically do it on chat GBT. So I think this is more of a market or or a market norm that you just uh want something that's specific. Also in terms of LLMs, I think if it's very niche and very focused as we said like the knowledge layer and the rag model then you probably will get a better performance in one specific area just like I know that's extremely different but claude is is for instance better in in coding for I'm sure there's a reason uh that's very quantifiable right uh so yeah that's that's why you're going to see more and more focus and then the pricing point part is so these tools, they make money on the B2B more than they make on the B2C. So, they're going to and again, in my

opinion, they're going to allow the B2B to spend all the marketing and the reach and the accessibility and then they they would just be selling to the to the other tools rather than trying to reach me or you or someone just asking I don't know how many >> they're trying to go more B2B. Okay, that's an interesting thought. Yeah. No, I that's that's definitely and and do you think this is going to like push out dayto-day um consumers from using the product or do you think they're still going to be available these do you think these products are still going to be available like these LLMs in general like CHP etc. >> Yeah, I think they are but >> they're loss leaders, right? Is are you saying they're loss leaders or something? Yeah, but then they're gonna jump the price up and then all the

other tools uh like for example if I as a as an entity pay $200 a month for chat GPT then I can offer fractional or pay as you go model and this this was the same thing happened in Belcom bubble and then the hosting and then sharing. So yeah was I just think Chad GPT is not going to be for $20 in in in in a year or so. Maybe it will 10x that and then consumers will just back off and look for other tools that pay page also or open AI, but they give you a very uh focus tool. >> Well, I mean just some data to back up what you're saying. Apparently the pro subscription, the $200 per month subscription that Chat GBT is charging, they claim uh they're losing money on it. >> Yeah. >> Right. So like >> something wrong with

that. >> They hit 10 billion in revenue, but they're still living losing millions of dollars. Yeah. >> Yeah. It's an interesting. >> And then usually like it's a marketing approach where you just not marketing or just you you let people get used to it and then and then well I I I personally can can't live without an AI bot in my in my phone at the moment. Like it's just weird googling stuff now. Uh even Google when you Google something Gemini is out up there. If it's a it's a >> Gemini is the default. Well, what I find interesting about I guess what's your thoughts on the the Gemini component to it? Because I've actually found in the last few months Gemini is a more interesting it's in an interesting position here because Google is making money handover fist. It has been for a long

time, right? This is essentially just a part of their company at this point. This like a tech it's like a a tech part of their company. Um what I mean by that is like internal tech. So what and it's an enhancement of their current product. I think personally OpenAI is in an intriguing position because they're losing billions, millions, and some are claiming billions. I I don't know if it's quite that much. I'm not sure. But Google's fine with this. They're just putting their devs on a new project, for example, that isn't as directly correlative to money. Sure, it's losing money in the context of that department, but their company is still, you know, extremely profitable. Do you think this is kind of putting OpenAI in a weird position because they're essentially um trying to get to critical mass when Google's just like sitting there like,

"Okay, good luck." And they're competing, but they're not actually having the same deadline issue because at some point Microsoft will stop giving Open AAI money. >> Well, okay. So just uh comparing Chad GPT and and and and Gemini uh I honestly think in in a lot of areas Gemini is is better or has was more convenient let's say. However GPT has really mastered the the human part. Uh when you when you speak to the agent or or the NLM you you really feel like you're speaking to someone right and and you you really vent if that makes sense. uh with Gemini is very it's still robotic in terms of how it speaks and I think this is the the what's differentiating uh both in my opinion also like trap was the first to market and then it takes it takes a lot of time to

to change that uh like even now usually it was Google Google stuff right and then you you say chad stuff even if you're going to use cloth so it's in everyone's head at the moment it's going to take time to push them away but just so yeah go back to what you said. Uh I honestly think Google is more uh let's say scared than uh we think and maybe that's a very big accusation from from someone like me but uh I think they saw what happened between Tik Tok and YouTube uh and then a lot of their products are just crashing. I think Google cloud is not the optimal cloud people go to, right? So I don't know like big companies fail at the end of the day. Sometimes it takes a long time to fail but they do and this is what happened to

Yahoo for instance. >> Sure. Yeah. >> So I I do think Google is panicking a bit because a lot of their products are just getting kicked out of the of the the Gen Z norm or like the normal usage. uh however I do agree that they they they have a longer uh let's say uh runway or breath or they they can they can handle more and also if they really push push that product up they have everyone's on Google calendar everyone's on uh Google meets like like we are like it's it's just uh it's everywhere so they have better odds I agree but I don't think they're they're there yet like tra is still in everyone's heads Yeah, it's interesting. So, I've actually seen some good improvements from Gemini recently, but I I could see what you're saying. I think for a while it was

pretty bad. I do think that. Um, but slowly but surely Gemini has kind of made made its waves. Um, in my opinion, you know, it's it's it's improved. So, it's just intriguing to me that, you know, Google's in this position where they have the ability to put money into it. They're not losing money. Um, but OpenAI is allegedly just burning. Uh yeah, V2 and three, they're just crazy. Insane. I don't think uh anyone can compete with that uh right now at least. >> Yeah, talk more about that. What are your thoughts on on the video generation? Because to me, I think chat was something I wasn't surprised could happen with AI, but I'm actually shocked by the capabilities of V3 right now. >> Yeah. Yeah. Honestly, it's uh every time I watch a video of V3, I'm just I can't take my eyes off the

screen. It's just uh and then it's also not perfect in a way that they they make it they make it in a way that it's it's humanized, right? Uh and I haven't used it yet because it's it's not allowed in Europe yet to year three or just I don't have access to it. So I'm I've I've been just watching uh what's happening out there online. Uh but I think even from a technical point I I have no clue how uh how something like that is is is built. But when you when you used to put something on Chad GPT, a picture or something or even Adobe or on Figma or whatever, the LLM processes images and videos completely different. It's it's probably binary or code based or like >> Sure. Yeah, it is. It I think it is binary. Yeah. >> Yeah. So, you couldn't

even put a picture of something and then ask Chad to edit it, right? Even when when I was building uh Trio at the beginning, I tried I tried to do a hack where where I can put a picture or agent that processes pictures and then you can take a picture of the schema uh and this would be like an easier token thing. But again, it didn't work. But I think something V3 did something extremely different where it actually processes uh images and videos just like we do. At least that's what it looks or seen from outside from the outside, right? You know, it's so intriguing to me because we're talking about images, we're talking about video, we're talking about text, and what is good and what isn't. And you mentioned as well big players, small players. You're trying to do something very specific. And my

question to you is it seems like you're pretty gung-ho on the fact that these big players are going to be the ones to do well in this opportunity. So, what what sets you apart and able in being able to do what you want to do with Trio? Because I think I think it's fair after hearing you be so gung-ho about the about the big uh the big players. I know you're in a data analytics specific kind of use case, but I'm curious. >> Yeah. And then there are massive players in the data analytics uh section also, but yeah. Uh but yeah just uh in terms of LLMs I think uh these LLMs will be making their money either way whether it's through me building trio or someone else using it directly right because I I am technically using them. So yeah, it's they don't have

the the let's say the motive to kick out small companies that use their uh APIs because this is their access tool, right? Uh but what differentiates let's say Trio I like to always give the example of uh Oracle and SCP versus UDO? UDO was a was a was a crazy amazing story for me because no one could compete with Oracle and SCP at at a time and I think UDO just created their own uh their own segment by focusing on very agile having a very agile product uh having it in open source at first to learn what's happening out there and then really giving support to let's say smaller entities smaller manufacturers and then like the ERP system is very agile as opposed to Oracle and SEP where they can it costs a lot of money, it takes a lot to set up and then

it's very it's a one-izefits-all situation and this is the differentiator in my opinion if I'm going to say it in a very high level uh high level perspective but all these these BI tools they they focus on a very specific uh segment or market and I don't think it's it's beneficial for them to tailor uh their model or their their product to smaller companies. It it doesn't it doesn't really make sense uh financially. So So yeah, even the the features, the persona, the everything we're tackling at Trio is is completely different. Even though if I just say a oneliner about my about Trio, it's it's probably going to be very similar to Metabase. But uh but yeah, I'm not sure if that had your question. >> No, it does. No, I was just curious. I mean and and where do you see where do you

see this company going in the next um 12 to 24 months? >> Yeah. So to be honest when when I ask myself this question I I I get a bit overwhelmed and and then I just focus on today right or at least this month. Uh but uh we are training our model not training we we we were first fine-tuning by the way and and this is a disaster. I don't think it's like for anyone out there finetuning is costs a lot of money and it's not very efficient but we do have a an Arabic uh side to it and it's not the translation Arabic of it uh but uh the agent can read Arabic schemas and this is this was much more difficult than I thought uh to implement and I I think expanding into the Middle East at the beginning would be uh would

be the direction or Egypt, Saudi Arabia, Dubai. Uh so this this region uh something that's that came up recently uh in in let's say Copenhagen or in Europe uh a lot of FMCGs uh they have IPs, right? So they have I don't know like uh like Pepsi or Coca-Cola. They're they're too big, but in general they have a certain part of their data where they can't put it in an LLM or an open source. And I think exploring having an on-prem LLM which has been uh coming up uh recently uh is also a direction I I I would like to explore within let's say next year. Um but yeah that's that's just thinking too big for now. >> Yeah. No, I'm just curious. I mean it's it's always um something that I think about when I'm you know I talk to people who are more

early. um say you weren't in this specific um I guess business so to speak. Was there any other ideas that you had considered in the AI space uh before getting into this one? Well, honestly, the businesses I I usually uh tend to lean on or think of are more operational because this is part of at least my experience in my edge is is having a more firefighting and operational rules. However, uh automation I think was was the the most recent thing that I wanted to explore. I found a lot out there, but having uh let's say a software house that's focused on automation was something I really wanted to do. Uh but then there's a lot of tools that can do that without a team. And then it was it was for me extremely competitive. Uh and I I can imagine that in one year

you won't need anyone to do that for you, right? Where even now with NA10 and Langflow and all these you can drag and drop. you don't even need technical people to to implement that. So, I'm glad I didn't. But but just on a personal side, I think automation is is the most thing that inspires me or stimulates me. >> H Okay. Um I really love I mean I really love automation. I'm I'm a huge fan. I I'm a big um make.com fan. Uh I want to get my stuff moved over from Make.com to nadn because it's cheaper uh to be honest. But, you know, it's uh what I'd really like to get into is self-hosting on NAN, but I I haven't I'm not there yet. Obviously, it takes time for for someone to learn things, and that has that's like on my low priority list,

but it's cool. Um, what's the one thing about AI that you're most nervous about? >> Honestly, it's my my kids. Well, I don't have kids yet, but it's it's just the future. Uh I feel like every every let's say relationship is is is now three people, right? Uh it's just uh the AI is just between every single person. So and this this this just proves that we're relying more and more on uh on AI which disconnects you uh in a way and then it it causes some anxiety like just our uh generation has more anxiety than the previous generation because we're just more aware. we have more access to uh more data, more questions, and it's just I think our brain is is is handling much more than it should, and it's only going to get worse. So, I I don't know what's what's going

to happen to kids in, let's say, 30 years, if not much less. Uh but I do think it it causes a lot of anxiety in uh in in society. Maybe that's uh too too deep uh for that specific podcast and focused on AI, but that's just my personal opinion. >> I mean, it's it's it's an important question, right? Like um I worry about my I don't have any either, but my kids too. You know, we have a lot of information coming in and the co-host of the co-host of this podcast has um kids and he they often use the chatbt voice um feature, right? Uh with the pro version and just being able to talk to them. I find that it's it's really inspiring what you're able to do with it. I think it's very cool, but also in the same respect, I'm like, man,

that's actually kind of that's actually kind of crazy where we've gone. Um, and I always ask that I I like to ask the question of what you're most nervous about. And I I like that you went with uh what's going to happen to the individual. Um, cuz some people could say things like how clawed for Opus when it tried to get shut down in that test by that consultant the other week that it it blackmailed people over email and that is concerning. But also, I mean, our kids. Yeah. I mean the next generation is the you know um steward stewards of the world you know >> I think losing control over AI like it could be like videos people can start I don't know imitating people instead if it's videos or pictures or voices there are a lot there's a lot that that could be

scary in terms of uh security right in a way but I don't think that's like the end of the world but when it impacts a whole generation and how like the transition how it's going to Uh, I think that's that's a bit more concerning, but ironically, I I do think that uh we at least as a as as people that are witnessing all that at the moment, not someone that's going to be born with all that. Uh we're going to value in person or physical or service much more than ever. Uh so it could it could actually play out the other way. Like you never know. But even for example, I went to this restaurant to where the robots serve uh well it's not robots, I mean just uh bots walking around your place, right? Uh and I honestly hated it. Like I I even

hate using even though like all I've been in tech for a long time, I hate ordering on apps. I I like calling the phone. I like going and >> I get it. No, that's that's fair, man. Like it's personal. It's personal. And I I think I think it was Mark Cuban who said recently that basically the um in-person sales people are going to to not sales but just like business owners and sales are going to thrive because I I I've noticed it what's really intriguing I used to be more like maybe introverted and nervous when I was younger and then as I've gotten older I like do this kind of talking stuff and I you just get more confidence the more public speaking you do and it's so odd now I have more like like in person haven't met them speaking capabilities than like random

people who are in their 50s. And I'm like that's backwards. I'm like it's even it's even reaching higher generations. And I thought it was mainly going to impact mine which it is in droves. I'm 27 but I think just the the habit of getting into I've never really communicated with you before before this. We're having an entire conversation. It's it's like natural to me at this point. But it's like that's something that we we're going to lose if uh we continue to go down the route of um dig uh digitizing everything. Um and uh convenience without humanity is a little bit of a a road that I would prefer to not go down uh after there's diminishing returns. You know, I'm fine with Uber Eats or whatever, but like um being able to say, "Hey, thank you so much for driving all the way out

here with the food." Not a thing anymore. So go to trio.ai AI and uh check out trio.ai and we will make sure that um everyone's going there and if they're going there, we got to be ready. July 22nd's the the alleged release date. So, we're really excited um for that and we're really appreciative for having you on. So, just wanted to say to everyone who did listen to this episode. Like I said, check out Triaii and also leave us a like on YouTube or review on Apple Podcast. Uh ask any questions about the product that you're interested in and uh we'll see you in the next one. Peace.