The Ultimate Offline AI Secret

Then in the last, I think 15 years, everybody talking cloud. If you say you are doing on-premise, you're going to say, "Are you crazy?" But actually, that shifted now, specifically because of this geopolitic thing. We had a lot of customers say, "I like your stuff," but I cannot afford to run this within AWS or Microsoft. I'm Jake Aaron Deloreal, born and raised in Silicon Valley. They're here to take you behind the scenes to show what it's like to be a startup founder. You're sure they're on problems they face. The products they build, an effort to make our lives better. I'm excited to have all this today. Nama Bach, the co-founder and CEO of Understand Tech. Nama, welcome to the show. Thank you so much, Jake. Happy to be here. Well, happy to have you. Before we jump in, where are you joining us from today? From South of France, a city called Montpellier, very far from California, but almost the same weather. Well, I know it's a great country. It's a great place to be. There's a lot of people that come to Silicon Valley from all over the world, and I go back between Laguna Beach and Silicon Valley. There's a lot of French and a lot of people from Europe that come to Laguna Beach all the time. So pretty well-versed with that. Back and forth, diversity and international traveling. It's really fun to be here and also to be where you're at. Before we jump in about your current company, Understand Tech, tell me a little bit about yourself. How did you get into technology? What was the inspiration to jump in? We'll talk a little bit about your company, but what was the path that led you to starting Understand Tech as well? Sure. So you said you were born in California, or raised in California. I was actually born and raised in sort of Morocco in a small city called Daqla. It's a very new city for Kaik Surfen. Yes. There is no tech there. There is no big Silicon Valley company. It's only a small city people doing surf, I think. I was born there, raised it. Then, a bit of the 2000, I came to France to do my engineering school. After my engineering school in computer science, I have done some business degree, and then I joined Tech Company Orange and ADME and France. This is what I was doing research on cyber security and stuff like this. And then I joined the company, which is your parent company, but it's more American culture called NXP semiconductor. And this is where my career really started to begin. And they have halt multiple roles from product management, product marketing, business development, and they learned a lot about the US, because there was covering America's region for five years. And this is really where I get most of my knowledge and business, network, and this is where where NXTEC come from as well. When I was at NXP doing some e-conductor, those companies they have a lot of products. It could be thousands of products, right? And that was taken more than 40 percent of my time was traveling sometimes to the US, to teach people by customers something that exists inside a document. The data sheets, it's an up-note, it's a source code from something. And the only way to scale that knowledge was to hire more, but not everybody can, not every company can hire on hire. And so that knowledge that they developed across MFTP, it was very hard to scale it even if we have done training to set speed people. And this is where the idea from NXTEC came in. And I was doing some AI on the site on the weekend. And AI typically knew from the start that it's called solved that problem of knowledge that some people have. Now with Virtueard subject matter experts powered by Adelaim, you can solve those problems. And I just decided to go with a friend of mine who was my biggest customer, by the way, when I was at NXP. He's based in Philly. We decided to go and create the company. And this is how it started. The company has been two years now. We are more than 10 people. And we just achieved it 500K AIR, just this quarter, with big name. Just after one year's product launch. And so we are really scaling a lot on this year. We believe we're going to close the year by more than million AIR. Wow. That's it, that's. Yeah, it's exciting. It's exciting time for a lot of companies, a lot of product visionaries that can bring ideas to market relatively quickly and test it and see if there's a market for it. Sounds like you're finding your market and in revenue, which is exciting to hear. Talk about your background a little bit. You said semiconductor and you talked about reviewing content or training people up on. Maybe what documents are saying or what they actually mean. At one point, you said that Genai could replace a lot of different types of people and different types of skills. What is your product doing now? Give us a use case of where it's helping an organization, $500,000 is a lot. If you think about it, for some people and for others, it's not that much. What is it that you're solving for that customer today that's helping them realize an efficiency or an accelerating something that they couldn't do before? Sure, let's come back to that same semiconductor industry. So the problem we solve, which is very hard is distribution. So when you're at a hardware company and you build hardware, you have two channels. You can sell direct and there you have an account manager, you have a support dedicated resources with each account. That's fine. There are some problems, but that's fine. And then half of the business of marketing of those companies is done through mass markets. Meaning customer that you have no clue who they are. They are serviced by distributor. Distributor like things like Avanet, Aero, Futures. There are a lot of them there. They are massive. They distribute a lot of products. But those customers typically too, so those distributor to sell your product, the information need to be easily accessible. I need to access easily to your data sheet. I need to understand your product. I need to be able to compare product to compare your product against your competition. I need to be able to get a presentation for a specific customer, like let's say I'm seeing Bosch or Honeywell. I need to build a lot of it. And this is where the things start to be bad, because that process is a mess. You can't have resources available 100% of the time just to answer the distribution question. So they built some setup to do this, but that's a huge problem. And the one who solved this problem, meaning the semiconductor, it's the one that went really from revenue. And so we solve this by creating what we call sales channel enablement portal. It's a portal, all powered by a UI. So we're going to put all the products of the company behind them. We have LLM. So the distributor come ask question just using natural language. Prepare slide. I'm going to see Honeywell. Can you give me a presentation? Boom. The UI is going to give you a fully tailored presentation to Honeywell with the product from that customer. OK. Now I want to understand for this Honeywell device what is the tear down. So when in the chip that they are inside, I can replace them with new one. The tour is going to give you this by going and doing some deep search with multiple agents that we have inside the internet looking to the FCC documents stuff like this. So a lot of the two that's going to boost the sales process and have the distributor to access easily to the semiconductor documents. And so at the end we see impact on the revenue directly from those semiconductor company. So this is just one of the example. Definitely the baseline is our platform, which has RG and LLM. So it has a lot of things. But this is not what the customer they care about. What they care is really this application layer that we provide. Got it. So I know we talked about the semiconductor industry and you give a great example of that. I could see a lot of value as a distributor. You might have multiple products, multiple vendors, multiple sales calls you're on. And you need to have the information you need. And you know, to get in front of a customer you want to be prepared. So you can give them that content that they can present and it can help accelerate a sales cycle or close a deal. I could see that's real tangible value for a distributor. Is that your ideal customer or are you focused in other markets as well? Yeah, good question. So not only because I come from that world, they know more problem to solve versus a standard guy. So yeah, it's our primary market. But then definitely we have three other markets. Number two, it's standard organization that's another mess. So we are talking about millions of documents. Each of them is more complex than the other one. And so same thing we we address them with a lot of tools. And we work with biggest, biggest standard organization. The third market is the device manufacturer. Still on the hardware, but now it's not the semiconductor, not the chip inside the device is the complete device itself. You know, they have similar problem, but there it's more engineering productivity. So we have tools that are low to do test case generation. So from a speck of product, you we address that speck. And then we give you the test cases in one hour. This process, you know how much it takes Jake today. If you want to do it with engineers, sometimes 12 months with multiple people. You know how much it takes to do it in our side hours. Wow. So we break down that complete process. And this is something we built from scratch with organization. We learned with them the problem. And we enhance it this with a lot of iteration in mind. And at the end, now you have a tool. You put your speck, you have a requirement, you have test cases, you have the script, and you just go and execute. And nobody want to do the test cases, by the way, because it's very painful. Now you have this that can do it with you. So yeah, that's the market. And then the fourth market, which is mainly regulated. And the street, so accounting, we have some accounting customer HR. Because those they care a lot about security. And they cannot just go and throw up their documents. I don't drop a corrup on I. They need control. They need security. And that's one of our value proposition. Wow. You got multiple angles to your platform based on different use cases and different customers. So really cool. How do you manage the focus on who to target and how to do the outrage? Because to get above the noise of a product is one thing in this new world of AI. But to go into four different categories, really, or channels, rather, you're taking on a lot for a small team. What's working for you? A good question, but just to go back to what-- you said it's multiple angle you're right, but there are still two common things that we repeat. Control and application layer. That's what we do. The control is the baseline. Security, sovereignty with the platform. Everything can run air gap, it can run in a box. I showed you this last time. Combined with the software layer, which is tied to every customers, then that's what gets our success. How we do it? How we scale? Because you're right, we are just about 10 people. You can't compete with the race of somebody like control pick of an ii. So we started at the beginning and here I'm telling you the garage thing. We started to build up manually. Like for every customer, we're going to see their problem, we're going to build this up at other customers' problem. With one condition, I'm a customer, there's going to be very cheap for you, but you're going to allow me to set the same app to another customer. So this is where the repeat motion came. So we invest once with one customer, they definitely pay some unrea to help us do this, and they take the same app and the repeat at other customers. And because you make it with one customer, you understand the problem, you understand the scale issue, you understand the distribution, you go to the other ones, they say, "Hey, I have your competitor using it, and then I can hit the app really to go." So that's what we have done to start, but now we are going to more a real scale motion. We developed a tool that's a tool to build up from scratch, just using a prompt. You come as Drake, you say, "Hey, I want to build this app that does this and this, that's connected to this system." I think you should know about Lovable and there are a lot of, so we build that internally that we are using to accelerate our own work with one specific, everything should always run ergaptab. The tool can't show run ergaptab and should not rely on any public at a time, so we need to be able to run it within a box, within my own promise, within my VPC. And this was what's a low-ass, definitely to scale to thousands of customers, much easier. Wow, talk about that a little bit, AI in a box. Is that something that is, I don't know, what was the strategy behind that? Because today, a lot of companies are building, they're connecting to the big LLM, is there in the cloud? Is your product or platform specific to, like, within the enterprise of the customer you're working with? Does it actually go out to the cloud as well? I mean, I'm not an engineer, but maybe just walk me through that a little bit. So let me show it to you, by the way. I have one here, I think. Everybody should know this box, so we just put a sticker on it. But if you are watching Nvidia, this is coming from Nvidia directly. So Jensen made us a lot, a huge gift when he released this one last year, because simply the market shifting clearly to on-promise now with LLM. Didn't the last, I think, 15 years, everybody talking cloud. If you say you are doing on-promise, you're going to say, are you crazy? But actually that shifted now, specifically because of this geopolitics thing, Europe, one, sovereignty, US one of the, and so we had a lot of customers say, I like your stuff, but I cannot afford to run this within AWS or Microsoft. I don't have DevOps team, I don't have resources to run them. This, do you have a solution for me? At that time, Jensen from Nvidia, they announced this box with even mosques, said them, this is exactly what we need. We take this box, which was meant for a lab. This is how they marketed at the beginning, and we customized it from software perspective. And then I said this as a real server, where you can have 50 people connected to it. So now I can go to an accountant company, I can go to a small other team within a big company, hey, you can do on-prem without investing million dollars. You invest a few tens, this is typically between 15k and 2k with LLM with everything inside, just with box. Policeman electricity connected to your internet and done. Managed by us from software perspective, no DevOps required maintenance, security, everything is done by InterSentek, and it's actually Nvidia law that we were concerned that the beginning that Nvidia say, hey, you're not allowed to do this, but they saw us in an event and stuff like that, and no, actually you are doing good job because we are creating value on top of their box, and now we are trying to work with them to do some call marketing together. Wow, that's really cool. Hold that box up again for the listeners driving down the road, they can't see it. This is about the size of like a. My hand, Matt. Yeah, like your hand, like a Mac mini, it just looks like something that you can. Much better than a Mac mini Mac. You can plug into the wall, you're up and running, that's incredible. It has, sorry, it turned out there, but the beauty that Nvidia made, and again, this is not something we invented, they created this connector, 200 gigabytes per second, you just put this connector here, and then you add another box, and then from 50 user, now you are going to 100 user, and you can do this up to three box, you know, so you have a real hack. Yeah, that's great. Technically, who are you selling to in an organization? Now, is it the engineering teams, is it anybody that's operating an organization? Who are you selling to specifically in a company? Even the box or the. I understand, like, solution globally. Yeah, when it comes to the box and the solution, probably I would assume would be the business users, but from the box, the technical side, are you selling, is that going to the CTO or like. Who's interested in that about it? So, it depends. So, this has a lot of success with SMBs. So, SMBs is going to be the IT team, where they are looking for a sovereign, from my solution, but they cannot afford to buy a huge Dell or HP rack, you know, big, big machines. That's how it's all gets. Then within the larger, large company that dip into each team. So, IT can adjust qualified, but the people who can really drive the man's. People who are doing, for example, patents, when you are doing patents and you want to use AI, it has a huge value on search, on writing, but the day that you use charge IPT, you lose the patent, because the confidentiality is the heart of a patent. And so, the patent team called Drive Defending the Man for us, coding, engineering team who are coding, as, you know, everybody use Cloud Code, they use cursor, a lot of tools, but there is very expensive, because it's token-based, you pay in function of token. They use the box, because the box we don't charge part token. It's a subscription for six, six, six subscription, and you can wire like, like, fire, million of token, you're going to pay exactly the same thing. So, it's more from cost reason than efficiency, because the box has a GPU, 120 gigabyte for people who are interested, and you can have multiple reasons, and you can just go, similar way you are doing with Cloud Code, you do it locally, securely, with no cost in terms of number of line of code, you are right. Wow, that's amazing. So, no tokens, no extra charges, you're not surprised by at the end of the month when your engineers are reasoning and developing and testing, and you're seeing these bills that start to rise, you didn't account for. This is really controlling your cost, at the same time, being able to continue to stay at pace and innovate in AI. I'm security exactly, with security on top of that. So, whether you're in an industry that security is important, or an industry that's really not that important, it's still a good solution to look out from just a cost perspective. Yeah, absolutely. I mean, even large company, we did not believe that the beginning of the world would be a market, but actually there is, because just the head ex of getting IT to approve your system is so huge, you know, you have all the documents. But when you come with a box, and it's fully ergap, then you remove a lot of those sliders. Yeah. So, we tell the team, even internally, to go faster on adopting a AI, because those sliders are a company, even if they know AI is very, very useful. The policy is the processes killed it. Sometimes you spend over 18 months before you get in. I'm not talking about selecting your solution just to get qualified with procurement, with legal, with a lot of blah, blah, a lot of processes, and the box, I'll say, that process, because everything is ergap. Yeah. What's the biggest challenge for the company today, where you're at, at the stage you're in? Challenge is always resources, Jake, that's, I think, a common problem. As a leader, typically, you need to balance of growth and higher, right? You can grow a lot. You may go and hire a lot, but, you know, as every growth, there is a lot of downside. So, and we try to balance, and so we try to do with less, so many, we have more than what we can handle, and it's really resources, definitely time. That's the biggest problem. And I think a lot of startup, this will do, isn't it, to them as well? It's resources on time. Yeah. Well, you know, we're hearing so much about on the resourcing side and the people side, it's one of the biggest challenges and the biggest opportunities for all companies. No company exists without great people identifying who they are, and bringing them on board at the right time is always the challenge, but for the companies that do it well, they're seeing a lot of growth and success right now. And for you, having had other startups and been in, you know, technology for a long time, talk to me about what's worked for you when it comes to hiring, because you've got 10 people, I'm assuming they're good, and you're building on top of that. What's working for you in terms of identifying and hiring the right people? What's your strategy? Well, first of all, I will not say I had no clue, but I had some some perception about what is the good profile, but actually turned out it's completely false and wrong on what they had in mind and what's the reality. And I'm lucky that we haven't met a big mistake on the on there was maybe one, but in the 10s, they are they are rocks, rocks down without them. The company will not be in here. So, I mean, yeah, and let me know if I don't answer the question there, but the really the way we have done it, and this is systematic now, we we try. So there is the technical aspect, so the resume of the person, which has done before, but actually the feeling it's work more than the technical background for us. For example, some of my dude here, I can tell you there is no company that would hire them from the first interview, just the way they speak, the way they they they they they are wearing, you know, now I'm talking about the standard company, but to now they are the best people actually. We spend a lot of time just trying to understand if this is a good fit for the company from behavior perspective, from mentality perspective rather than technical, which, which anyway, even if somebody does not do something, as long as they have the good mentality, they can learn. And that was our driver, which we haven't really set with my co-founder. It was just doing it on the fly, where not to be company with processes. They turned the thing that helped us to keep most of people that we have hired. We have only one intern that we stopper to work with, but everybody who joined the company since 12 months, they are still here and they are super happy. Yeah, that's great. Talk about the one that didn't work out. What did you miss? Well, I'm going to tell you that you're going to maybe laugh, but the feeling was okay, the resume was okay, but unfortunately the person, haven't had enough experience to use AI tool in a better way. Let's see, let's see, this way. So in other words, you throw up something like cloud code or or open AI into the hand of a young, a young engineer who has no experience. You can create a huge mess. And that's what happened, typically. So creating code without review, pushing things to, you know, production system, without review, even the review we have done it, but we were two at that time, just three people. You can't afford to do a real review like what you are doing. And so that that was the biggest issue. And we tried to fix it, but you know, that it didn't work out. That's what that's as well and learning. So we established processes we go now every single line of code, whatever you do with AI because we're an AI and a company and everybody have to master use AI, but every single line of code that's going to production is verified and another review would buy all of people. Really cool. What's the hardest early hire you've made so far? Early hire, you said? Yeah. Like early on. I know you're still early. You've only, you know, 10 to the boys now, but what was your hardest early hire? I think she's going to see the video Clara one of her. So she's our rock star marketing. She's my right hand on everything they do in business. I know. I know. I Clara. I don't know what I can say, what I cannot say because this is going to be quick, but I'm going to, I'm going to say the story, Clara. So when I met her, she was still, she just finished school and she was looking for a job. My wife put me in contact with her and then, okay, let's go with us to this event. That was the first meeting. I don't know her report. Hey, you know what? Let's meet in an event. And my co-founder was here. And I think she believed I'm crazy or something. They do. He doesn't know me and he's enlightening me now to go for a two-day three-point event in Khan, South of France. But it turned to be the best test you can have. I see in her, but customers, I see in her how she can adapt and learn very quickly. And then different after the two days, we can see there is a clear fit between her and the company, what we want to build. But we didn't have money to hire her at that time. And she was just a contractor. And then when we have money six months after, she was already on another job. Oh, no, I can't move. I just, I'm happy here. And man, I struggled for three months. And Clara, I think you know the story. And then one day, she said, okay, let's have dinner and simply have a discussion. And I think she understood what we are going to. And she simply quit and joined us. And she is my best hire as of now. But definitely I rock star with the team from energy from more perspective. Yeah, really cool. I love those stories. You know, I've interviewed 30,000 people and all sorts of different interviews start and end, but they're all a little different. What we've found. And this is, I want to share this with anybody, is that the companies that we've worked with and we're in the people business helping them build and grow is the ones that have scaled and done it well. They've hired right. And what they've done, it's really two things they got right. One of them was they were clear about who they were. Their identity, their culture and culture can mean a lot of things to a lot of companies. But part of it is how you operate and how you think. And in this world of AI, are you a systems thinker? Are you a problem solver? Are you an engine builder? Are you a creator? And can you do the job? So kind of aligning your culture with what you need them to be like is kind of the first step, the second step and the last step is to make sure that they have the competency to do the job that you're hiring for. And if you can get those two things right, 90% of the time, it's a fit and those companies scale. Everything else from our position is kind of noise. You do the interviews to make sure that you get clarity. Do they can you trust them? What you're got to tell you? Are they saying what they actually have done? You do all the questions too. And you check the references. If you can just kind of keep a focus on those two items, you're going to do well. Lots of companies that we've seen have scaled significantly. And those are the two things they focused on. So that's just what we see. What you're doing, it sounds like you're you're doing a great job. You're off to a good start. You've got you know, a team in place. It's next 12 to 24 months look like for you in terms of growth. And if there's roles that you want to talk about, feel free. If anybody wants to find you, you know, it's no cost to anybody. It's just you connected with candidate. So talk about the kind of roles you want to build to close the gap as you grow. So let me talk just about the very first. I've then definitely read an opening. And I'm happy that you're sick. We can talk about it. So while everybody is fighting today on a yeah on the infrastructure, Lyre, we clearly understood based on direct customer conversation. I mean direct customer signal that the fight is not on the LLM Lyre. It's not done. It's made on the application Lyre on the software that come on top of it. Specifically, after four years or five years now that chat GPT exists, everybody now understand how to take an LLM, execute it, run it, you know, chat, what stuff like this. But what enterprise need I'm not talking about consumer, they need tired application that connect to their knowledge, to their processes and solve real problems for their sales team, for their engineering team, for their product team, legal team. And that's what understand tech is doing now. And we are one of the first that is attacking this market. We want to be typically in 15 months, what one years I have and my head was thinking French 18 months, the 365 up store, you know, when you connect to 365, you have PowerPoint, Excel, you have, you know, every function has something. And the centech, that's what we are building. An up store of business app where every enterprise function will find something that's going to boost their productivity. Okay. We already started today, we have more than six app, they are used widely by the end of the year, we want to have 30 app, and then next year actually we have even a marketplace. So that's what we are going to and to execute this, definitely we're going to need resources. And today one, it's, we are looking to a system architect. So somebody who is really like a geek who know from 10 back and LLM, he is typically a horizontal. And the second role we are looking for is a technical business development. I'm spending a lot of my time, the wind quotation, working with customer to understand their problem, suggesting app, and that's putting a lot of my time, right? And I'm only myself. So we need somebody who can typically help me there and be my right hand on this aspect, on semiconductor industry and standard. So he needs to have knowledge on that market. Really cool. That's great. Talk about location. Do you want them to be in France? Do you want them to be in the US? Where do they need to be? That's a difficult question. If you ask me, I want them to be here, close to me with the team. But I know that's not all the time possible. So we favor, by the way, physical location, because when you are a startup, we have whiteboard everywhere. And that you cannot replace it with whatever remote tools you have. But for the BD role, the technical role, differently here, for the business development role, it could be exception that somebody could be in the US, whatever in the US, because that's anyway BD. So whatever. So there it's a remote is acceptable, but for the technical role, definitely in the office here in South of France. And I will not try to sell you South of France. It's one of the best places in the world. I believe that. Really cool. That's awesome. Nama, really cool to hear your story. I like that AI in the box where everything secure, you can plug and play. It's up. It's running. And there's not a lot of security issues. So those are, I think things that I'll remember and I'll share with, you know, our community as well. In terms of if anyone wants to find, understand tech, where do they go? Understand. Tech. Understand. Tech. There you go. I like it. Well, Nama, thanks for coming in and sharing your story. Appreciate you spending your time with us and to our listeners. It means a lot to me. You spent your time with us today. Our community is growing significantly. We've almost 100% in the last six months, which is really fun to see. So thank you for listening and your replies and the community. I'm your host, Jake, Aaron Villarreal, Sonning off for now. I can't wait to catch up with you all in the next episode. Until then, Nama, the world. Take care. Thank you. Take care. If you like what we're doing, don't forget to subscribe. Leave a review on Apple podcasts or wherever you listen and follow us on YouTube where we go behind the scenes to learn what it takes to be a startup founder. [BLANK_AUDIO]