Leonardo Ubbiali: Founder & CEO at Visum Labs

0:00Leonardo Ubbiali:Hello, welcome back. I'm here with Leonardo. Leonardo, please can you give us a super quick introduction, who you are, what you're building? Hey Seb, thanks for having me. So I'm Leo, I'm building Visual Labs, it's an AI and data consulting company. We help business implement AI solution end-to-end, from building strong data foundations to AI engineering to also product engineering, so that we can deliver the full solution. our focus is like driving P &L impact so we are very keen in like always showing like our clients before we start the project what might be the impact that might have either on increasing revenue or decreasing costs.

0:37Leonardo Ubbiali:And why is it that this is the business that you've decided to build this is not your first business you know what is it about this problem this area that you feel like needs to be solved? Yeah so a couple of years ago I went through YC with my previous company and I saw from there basically the AI revolution starting. I came back to London and I felt how a lot of businesses were trying to implement AI into their workflows, into their processes, into their products without solid data foundations. I do have a, let's say, data scientist background and I could see how without strong data foundation, it's very hard to build an AI solution that is reliable, that is production ready.

1:22that they can deliver the promise that they usually come up with. And I saw this opportunity to bring together, like, let's say the old data capability. So data engineering, analytics engineering, together with the new kind of like AI, LLM implementations, right? So I saw this opportunity to bring together these two services and help other businesses bring like a real impact into their P &L.

1:47Leonardo Ubbiali:Amazing. And you only announced very recently. What was that reaction like when you kind of announced the launch of your new venture? Yeah, we got a lot of interest. I think a lot of people are trying to figure out exactly like what we do, which has been also good feedback. Like we are definitely a service company, not a product company. And our goal is to help other businesses implement AI into their processes, into their products. And I think that's a clear distinction that is needed in the market where a lot of companies are product companies, but they also do services. It's not yet clear. We don't have any platform or any product as of yet.

2:27We just help other businesses implement AI on the shoulders of giants. So we are pretty like provider agnostics. We do have a few partnerships, but when it comes to what LLM we choose for a given project, we just assess what's the best for the use case at hand, for example.

2:45Leonardo Ubbiali:And can you give some news examples of like the types of things that you are looking to implement? Yeah, sure. Like our services span from like the AI agents, like both in voice or like AI chatbots. Like, for example, for a manufacturing company, we build a custom AI chatbot that helps the operators of their machines understand how to use the machine by building a rug on top of like the user manual of those machines, like hundreds, thousands of pages. that's one classic example other example might be voice ai agents to handle calls like for restaurants and things like that uh all the way to more like let's say data engineering jobs by building like the data warehouse for a huge fintech that needed some help to set the data foundation to then build uh ai use cases on top of that both aml or underwriting models that we can take all together in the next phase of the projects super interesting so we're seeing a lot of companies at the moment build and deploy AI in these use cases, but I guess from a product perspective.

3:47Leonardo Ubbiali:So build customer service agents or voice agents and then sell it to companies. The approach that you're taking is helping companies to build and implement their own. Is that right? Yeah, that's correct. The way I see it is that either you have a very specific workflow that fits one of the off-the-shelf products that are available on the market, and in that case is a great outcome right you can just like pick a product implement it into your workflow and that works maybe that's the case for like ai customer support for example i think there are great companies in that space and where visual labs can help that is basically help you assess which is the best one but there are a lot of cases where your workflow is not as simple as a product company might wish for right so there is no fit between your specific need and the products available in the market and then that's where we can help you basically build a custom solution if it needs to be amazing so you are a services business are you using any ai tools your own workflows your own agents to help you optimize and deploy your services in a way that's going to be more efficient than like typical consulting or services players yeah 100 percent we want to have a way higher margin than a typical kind of consulting business and to achieve that we must implement ai in our own processes those spans from let's say proposal building or assessment of the projects which is very very streamlined like using like llms and other tools yeah all the way to implementation.

5:26So for some services where I say the degrees of freedom are low, so they're pretty standards, we can use AI to streamline those services on our side, right? So the cost and the speed, the speed goes up, right? We are pretty fast in delivering it. And the cost of serving the marginal client goes down because we have standardized the process, and therefore we can leverage AI, not to fully automate it, but surely to streamline the delivery. and then once you have delivered a project are you is that you done are you you know would you

5:58Leonardo Ubbiali:partner with the company for six eight twelve weeks whatever and then you hand it all over and then you go your way or is there any sort of continuation it's a mix of the two i think it's like definitely a challenge of the business model like being very transparent uh a few projects are one and done projects yeah other projects are partnerships right where we set a roadmap with the client and we define basically a series of projects that we want to do together and those projects have a let's say sprint of implementation and then a recurring what we call continuous support engagement where we basically assist like their team to continuously improve or solve a bug on the implementation that we have done before yeah that makes sense and that i guess the whole is underpinned by data and that's true of all ai fundamentally um how are you able to help companies with their data is it about giving them structure is it about taking their existing data and using that and plugging it into models how are you actually using the data to kind of really drive value for an organization yeah most of times it's like helping them understand first of all that they need structure and processes around data.

7:11So a few companies come to us and say, oh, for example, I want to beat a customer-facing chatbot on top of our transaction data. These are classic FinTech financial services case where the user asks, oh, how much I spent on coffee in the last three weeks? And they think that just by plugging an LLM like ChatGPT on top of their database, this would work. That's very unlikely to work, right? So to do that, like you need to structure your data, you need to probably to build a data warehouse or a data lake, depending on what's the use case. And we help you with that. So that's a core data engineering job that is nothing new.

7:51But in this AI revolution, I think it's very undervalued. And more and more companies will crash into the world before realizing that like the data foundation is a must have pillar before implementing LLM. or generative AI for whatever use case they have in mind. So we help them with that. We help them bridging the gap between, let's say, the structured data and AI by building semantic layers or ontology, which are made famous by Palanty. Like Palanty has this whole like ontology layer, which I think will become more and more common in the industry, right? It helps like generative AI, it helps Chargipte, Cloud, understand how your data is structured and more and more companies will need to be able to build production-ready generative AI applications.

8:40Leonardo Ubbiali:Got it. And you are, you know, you're a data scientist. I think your co-founder is an AI engineer. I think you've got another AI engineer. So you, as a team, you've got a really strong data AI background. When you look at other consultancies or other service providers in this space, do you think they're always going to, or how well do you think they're able to adapt and help companies implement AI? I think we've seen posts about Accenture being the biggest player in making billions by AI implementation, despite arguably not knowing that much about it themselves. What's your take on that, the future of consulting versus your small, very focused, very experienced team?

9:22If you want to deliver an end-to-end service that spans from the strategy piece, so understanding what you need and why, to the design piece, which is let's design the solution based on your need, to the implementation piece. You need to have all of these skills, and most consulting companies I see popping up usually are very good at one of these, but not all of them, which is obviously very hard, right? What we are trying to do at Visual Labs is be very strong technically, right? So the last piece, the implementation piece, but also having the capabilities, the soft skills to guide the companies we work with in the top part of the funnel.

10:04I think it's easier to teach an AI engineer or a data scientist how to sell, how to be very close to the customer, how to service the customer. The other way around is way harder, right? So it's very hard to teach a management consultant AI or data. And we have this model at Visual Labs, which is not new. is like very famous now, forward deployed engineer. But what we mean by that is that the person we allocate to a project is in full ownership of the project end to end. We don't have and we won't have account managers, right? So the engineer is responsible of the relationship with the customers, is responsible of the communication with the customers, and owns the relationship end to end.

10:52And that's kind of a new, like, let's say, business model innovation we are trying to bring to the market.

10:57Leonardo Ubbiali:Amazing. no that i think it's a lot of sense um yeah yeah i can't imagine account managers or customer success managers becoming experts in like data infrastructure and ai anytime soon um you know i showed that on your website it looks like you've worked with a bunch of amazing clients already where's that come from has that been outreach personal networks relationships how's how's the selling process been of trying to like sell this proposition that you've built yeah like why reward in stealth most of the pipeline has been through my first or second layer of network yeah which i'm very lucky to have and now as we come out of stealth we are going to have a more let's say disciplined uh go-to-market strategy uh most of the channel will be probably content linkedin and also like some events that we are really focused on uh we're like pretty early like trying to learn also that like what's the best channel but definitely we want to come out of like my network and try to have a more let's say structural organic pipeline in the next few months yeah so that's interesting some of that stuff on linkedin is that going to be like building public showing case studies what's that going to look like yeah like we are about to release a few case studies like from the first projects we have worked on in the last like 10 months that's definitely one part of it and then also like showing like our capabilities like by i don't know sharing like white papers on how we think like a given use case should be taken from a given industry and other stuff like that that can be like the credibility around the brand so that then like customers can come in about nice i wanted to ask you know i interview ai founders a lot on this program actually not just ai founders but founders of all who are trying to implement ai or building ai products to sell to other founders people are always saying it's not about cost savings it's not about reducing headcount it's about efficiency it's about speed it's about the quality of customer service when you're speaking or you're selling your proposition to to founders to businesses what is it that they're looking to get out of what you could offer them is it better quality is it reduced headcount is it i don't know like is it efficiency is it effectiveness what are the things that make people turn to you yeah at the end of the day like from a business perspective, you are either trying to reduce costs or increase revenue, and any of other variables are just like leading indicators of that.

13:26So for example, if you are trying to be more efficient, you are trying to reduce costs, right? Because all other things being equal, if your team is more efficient, capacity frees up so that this team can do something else, and therefore you can have also this carryover effect on revenue as well, right so the the process we run when we meet these clients it's trying to understand like very clearly what they're trying to do it can be either increase revenue or reduce costs or both but we're trying to map out each project to the levers that then like have an impact on the p &l most of teams right now are trying to reduce costs that's what i see yeah although it's a harder sell than increase revenue so if you can share like we can show like that for example we can build a like a better underwriting model for a given fintech that at the same risk level can increase the number of approval of their credit card that's clearly like kind of like a revenue driver and that's an easy set right while on the other side if i build like some custom like ai workflows to automate how people do a given specific workflow that's a cost-saving kind of like driver and then other questions come up okay cool but then what i do with these people i cannot fire these people and it's also my job to help them understand how they can leverage those people

14:48Leonardo Ubbiali:in higher leverage jobs higher leverage tasks while leaving the ground kind of tasks like the very kind of nitty-gritty very boring always the same tasks to ai yeah yeah but are you seeing any sort of like cultural or people pushbacks in the organizations that you work with or are the organizations that you work with quite kind of forward focused anyway yeah it's very it's very clear across geographies and also across industries so we work in the us in the uk and europe yeah we have had clients across those three geographies and in the us like it's pretty clear okay cool this like helped me like reduce costs it's fine in europe for example on the other hand of the spectrum.

15:33It's like, oh, yeah, but then what do these people do, right? That's the question. So there is a little bit of risk, business transformation questions coming up. And also across industries, tech companies are pretty used to change, right? Change management in a way. They are not afraid of huge changes in their ways of working, and therefore they're faster in making their decisions. When I work with more traditional businesses, for example, like we have had like some engagements with manufacturing company companies i was referring to before they need like more assurance they need to understand more about what they're trying to do

16:09Leonardo Ubbiali:yeah okay and and for you and the vision of the company how big can you take this you know this is very different model and type of business to your you know your previous yc-backed company you know what is it that you want to build do you want to build the next mckinsey or do you think this is going to be always a small lead, highly effective, profitable business? Probably something in the middle. The company is profitable, is fully bootstrapped. We don't want to raise any capital because I don't think the business model also requires capital or is a feat also from the disease. So the idea is to grow organically year on year and then see what the options are as we move along, right?

16:51So we see like we set up goal for end of 2026. we will send a goal for end of 2027. And I think the company at that point might be at a sort of like bivocation of where they can go, right? Do we want to be like next McKinsey at that point? Let's see. Do we want to stay like a profitable bootstrap boutique consulting company? We will see at that time. But the fact that I own the company and we don't have any, let's say, outside pressure put us in the best place to make this decision and we see the market as we see the market growing and how different is has your experience

17:28Leonardo Ubbiali:been you know building a company without investors without the worry of fundraising versus you know your first swing where you were it was very much like investor fundraising focused is that a different like stress levels experience i just love to hear a bit about that it's very different it's a different type of stress for sure uh so you don't have like this kind of like you don't have like investor stress right so you own your company like you are totally in charge on the other side yeah you need to be profitable otherwise there is no business uh from day one you need to pay your employees to pay pay your service providers and make sure that at the end of the month the money that comes in is higher than money comes out and this shifts your mentality as an entrepreneur as a founder on how you deal like with some levels that you can pull right so for example now if i had like more money in the bank i would probably make a bigger investment on like the go-to-market function and how we grow but i cannot allow myself to do that and that's totally fine we can grow like at a slower pace and on the other side though like the upside is that i don't need to report to any investors and i can decide the pace of my company and i don't necessarily like coming back to the question before we don't necessarily need to become the next mckinsey for Visual Labs to be a great outcome for everyone who is involved.

18:50Leonardo Ubbiali:Amazing. Well, look, Leah, thank you so much for joining me. I think it's super interesting what you're building. I think we're going to see more and more businesses turn to people like you to get help. And I think we're going to see, I think we're seeing people who are struggling to just buy tools directly and just try and implement it without seeing the success. So I think it's really, really cool. And I wish you all the best of luck. Thank you, man. Thank you. Cool. Well, let's stay in touch. Amazing, man. Thank you so much.