932: Should You Build or Buy Your AI Solution? With Larissa Schneider

0:00Jon Krohn:It's the perennial question. Should you build or buy your AI solution? Well, what if you could have your cake and eat it too? Welcome to episode number 932 of the Super Data Science Podcast. I'm your host, Jon Krohn. Today's guest is Larissa Schneider, co-founder and COO of Unframe, a startup that has raised$50 million in venture capital to bring you AI solutions as fast as when you buy them, but also as tailored as when you build them yourself, with returns on investment in weeks instead of years. sound too good to be true? Larissa spills the beans on how she and her team do it in today's episode.

0:34Jon Krohn:Enjoy. Larissa, welcome to the Super Data Science Podcast. Great to have you on. Where are you calling in from? Thanks for having me, John. I'm in Berlin today. Excellent. And Berlin is the home of Unframe AI, isn't it? Well, we have a few homes, actually. We're kind of pretty global, you know, wherever the customers are officially headquartered in California. We have a pretty large R &D team in Israel and in Tel Aviv. We have our operations team in Berlin and then the go-to-market sales teams kind of between Florida, New York, London, all around. Tell us a bit more about OnFrame because it's a business model that I don't think I've seen before.

1:21Jon Krohn:And it seems like it's working really well for you. You recently raised, well, I guess the total of the raises, the venture capital raises you've done so far comes out to$50 million, including, I think, a relatively recent announcement. You can correct me on these timings and exact numbers, but this unique model that you have seems to be working out for you. So fill us in on what it is. Yeah, sounds good. Yeah, we started the company in roughly March last year, raised a seed round, then raised another round so our a round in march this year and um i think from day one we actually did everything against the book um so really not following the typical um playbook you know the if i go back to the very first vc pitches we did and we came up with this this crazy business model and everyone's like you know but you need to focus you can't start with like doing something for multiple personas and multiple industries and multiple products from day one and we said challenges we can.

2:16Because with AI, it's kind of everything has been reset. Like we were rethinking everything that we've done like the same way forever. And so we're doing it again and we're doing it better and more efficient. And we're really pushing the boundaries in that regard. So when we came out with our Out of Stealth announcement at the beginning of April this year, we actually came out with, we call it a managed AI delivery platform. And in very simple terms, we often actually refer to this metaphor of Lego bricks. So we built an AI platform that is made up of hundreds of different building blocks. So we looked at all of the most complex, the most challenging, the most time consuming problems that enterprise leaders face when building and deploying AI solutions.

3:02We packaged it and we use it hundreds of times over for all kinds of enterprise use cases. So someone gets something that is super tailored to their specific environment without having to prepay and or like no commitment, no cost involved until they actually feel business value. And, um, that's, that's what we came out with from day one. And yeah,

3:25Jon Krohn:it's been working well, no commitment, no cost involved until they feel business value. Yeah, absolutely. That's, that's how confident we feel about it. Um, and it's, it's funny. sometimes people are like you sure like a poc is no cost like yes it really is not because that's how we build the business and that's how efficient we made the platform and um it's really in tech unframed seems to be the only one doing it like that the comparable that chai my co-founder always mentions like imagine you're getting a new home and you want a sofa right like your custom sofa that fits your specific space and your style and your angles and whatnot your measurements well try to find a sofa builder that says, sure, I'll build it for you, totally custom to your measurements.

4:11And then you can try it. And if you like it, you'll pay me. Otherwise, no problem. I'll take it back for free. You won't find that, but at Unframe, you can.

4:17Jon Krohn:That is wild. And so then how do you know that they're not getting business value and not telling you? Well, yeah. I mean, that's always a challenging area, I would say, because what What we've seen a lot in AI specifically now is like there's been so much board level pressure, so much executive visibility on the topic of AI that a lot of people like, let's just execute on it. What can we do? What can we build? Let's just do something. And what we're really pushing for is for them to start with the ROI and the KPIs in mind. So what are you actually trying to achieve? Not just like which tech do you have at your fingertips that you could use?

4:56And so we really work, we call it like a business impact analysis that we do with the customers up front and say like, we want to build one or two or three different POCs with you, but let's try to find the one that actually moves the needle. And moving the needle for you means X. And if we hit that, then let's move to licensing.

5:15Jon Krohn:I see. I see. So you're kind of with them from the beginning on some metric that they're looking to hit with this particular feature or aspect of their product, their platform. And so it sounds like, correct me if I'm getting this completely wrong, but it sounds like Unframe is kind of mixing both services and SaaS kind of together. It sounds like you're able to have lots of different AI platform options for your clients that are kind of ready to go, but then you customize them. So there's some services, some adaptation to make the couch, say, fit perfectly into their space, be exactly the color and the fabric that they want.

6:01Jon Krohn:Okay, so it is a blend. It is a blend because we think that is very important right now because we've moved so far beyond this moment of generic software. It's like one size fits none. And so we really want to make sure that we offer that, but we don't charge for it. So all of our services and our AI product leaders that work on the specific tailoring of the solution, everything is included in our subscription. So you don't have any hidden costs, no additional charges that just pop up that you never planned on having. And now the subscription, that's got to be also bespoke. Like presumably some of your clients are using, you know, lots of functionality.

6:43Jon Krohn:They might add, you know, over time, a big client of yours might have lots of different pieces of functionality within their enterprise that depend on you. And so presumably there's different like kind of tiers of subscription. Yeah, we do. yes but we try to make it as simple as possible as well it's really fast it's all about simplicity we do t-shirt size pricing so depending on the complexity of your use case small medium large extra large but yeah we do it per solution per year and some of our customers as you say they started maybe with one or two use cases but now they realized how important unframe is for their strategy and now we've moved to like five six fifteen different type of solutions that they're running on Unframe at this stage.

7:27So they know how much they'll be paying.

7:29Jon Krohn:Cool, congrats. And it sounds like you co-founded this business, this novel, this completely new kind of business model based on the idea that we're now in a completely new kind of era because of AI. And so we're going to need to have, the businesses of the future are going to have different needs. The kinds of company that you can be building is different than ever before. And so let's try this completely new kind of business model, see how it goes. And you guys are crushing it. Congrats. Thanks. Yeah. Actually, you know, there was three of us or there is three of us as founders. And all three of us were working at No Name Security before, which was Shai, who's our CEO.

8:13He co-founded that business before. And we're just seeing this hype around AI everywhere. You know, like ChatGPT came in the world, changed all of our personal lives. But then we're like, how are we still working with business software that does not really give us the same experience and doesn't really, you know, there's some AI tools that everyone was trying, but it didn't, yeah, it didn't give that impact. And so we said, this is a good time or like the time to do it. And we jumped on it and have been building ever since.

8:43Jon Krohn:Excellent. Very cool. Are you able to go into maybe one or two case studies that would, you might not be able to because maybe everything that you do is proprietary for your customers, but it just occurs to me that, you know, we've kind of been speaking in hand wavy ways about, you know, it's, you have some, some, some platform offerings that are somewhat off the shelf, then they're customized. If you had one or two examples you could talk us through, we might be able to really visualize this. Yeah, totally. So there's three main areas that we work with customers on, and that's usually where we start with customers.

9:19The first one is all about observability and reporting. So think about BI analytics and search type use cases. I'll give you some examples for that. The other category is extraction and abstraction. So working with a lot of unstructured data. We're really enterprise focused. So as you can imagine, with some legacy, large global corporations, you have a lot of types of spreadsheets, financial statement reports, contracts, and so on. And then the other category are all of the AI workflows and automation areas. So all of the actions and decision making that can be assisted with AI. And let's say, for example, one of the top use cases we see a lot is around IT operations.

10:02So we work with one of the Fortune 10 insurers in the US and they approached us with their IT operations team having an observability department, 26 people. And all they do day in, day out is have like six tabs open on their screens with like ServiceNow, Jira, Confluence, Dynatrace, Splunk, like the whole stack, you know. And they're trying to manually correlate with these logs and all of the different tickets that they have. And they're like, that's not efficient anymore in the day of AI. So what we gave them is like a one pane of glass where they can see all of the correlation between the different tools made by AI.

10:41So for them, a lot less manual work, AI being able to see patterns at a much faster level, reducing time to resolution of all their tickets and just a lot more customer experience that they can offer. So that's a very common use case for us, as well as contract management. So we do a lot of that and think about contracts in real estate. It might be a lease agreement in HR businesses. It might be an employment contract or legal MSAs, all kinds of contracts where we abstract data because with many businesses that we see right now, they want to see the trend of their agreements, their expiration dates at scale, but everything just sits in a scanned PDF somewhere in a SharePoint folder.

11:23And so we come in and we take everything that they already have and make sure that we provide them insights and value of everything that's hidden there.

11:32Jon Krohn:Really cool. I'm crystal clear now on what you do. That was so helpful. And so does this mean that you end up being in a fair number of situations where a client comes to you and says, you know, we have this long list of needs that we think can be solved with AI. And you say, okay, from all of these, or maybe even, you know, some additional ones that you haven't thought of, these are the ones where we can make a big impact. You know, these are the ones where we have some kind of ready-made solutions ready to go. We can customize them just for you. But then there are these other needs that you have where that isn't our core expertise.

12:08Jon Krohn:Is that right? Yeah, that can happen, obviously. Like while we want to tackle a lot of problems, there are some great other companies out there and we don't have to do every single thing, even if we can. You know, we really focus on like the business business systems, the information type systems. So if someone comes to us and says, I don't want to pay for my Slack license anymore because it's gotten expensive, we're probably not going to build them another Slack, right? But there are actually a lot of use cases where we see a lot of repeatability. And that's where those Lego blocks, the building blocks of our platform come back in.

12:41So think about the first time we ever saw unstructured data, which was a lease agreement in commercial real estate. So think about like a 80, 90 page documents. We had not gotten that building block as part of our platform, but we built it. And that's where we built it so generically that now when a financial statement in Excel came in or a CV or an image type use case, we already have like 95 % of those capabilities. And we just need to change the last few and then put it into the library of building blocks that we have available and can use it again tomorrow and the day after for any other enterprise customer.

13:16Jon Krohn:So this partly explains to me why it's so important that you and your co-founder Shai are involved in founder-led sales still today, even though you've grown so much, even though you've raised$50 million, it sounds like one or both of you are still involved in some way or other with the sales cycle to all of your clients. And now I can see why that helps you identify, you know, what are the next Lego building blocks that Unframe need to build? Because it sounds like, oh, wow, you know, I actually three times this week, I heard a client say that there's this new kind of thing that they need. And so maybe that's a good place for us to move next.

13:52Yeah, absolutely. I mean, while we have an amazing team now and the company has grown like crazy, we're close to 100 people in the next couple of months here. So we really have an amazing team around the world. So we can't be everywhere all the time, but we try to be in as many as possible because for a lot of our customers now, as I said, we work with large enterprises. And sometimes when I meet people, like a few weeks ago, we had a conversation with, say, an AI leader of a top three Wall Street investment bank. And usually in my normal pitch, I always say, like, you probably have like 200 to 300 AI use cases mapped already.

14:28And she's like, I wish, Larissa. And I was like, what do you mean? I was like, I didn't know what to expect. And she was like, my backlog currently is 1 ,670 use cases that the business has brought to me that I have to execute on by the end of 2026. so I could have as many internal engineers as I possibly wanted. I will never get them done in-house so I need a scalable approach. And so that's how with our pricing model and the business value that we're showing them, the idea of Unframe is really not just like try to maximize returns on a single use case but them feeling value and coming back for use case number two and three and four and so on.

15:09And that's the partnership we want to build. We don't want to be just a vendor. We really want to be there to execute an AI strategy with our customers.

15:18Jon Krohn:On this podcast, I'm always going on about how Claude Code is mind-blowing, but now Claude Cowork is making my jaw drop as well. For example, I recently wanted to quantify how healthy my sales pipeline is for my AI consulting business. I simply asked Claude to estimate my sales for the coming quarter, and it brought info from relevant Google Sheets and my Gmail to create a professional spreadsheet of clients with estimated revenue for each one. Whoa, this might have taken me a day. Instead, it was done flawlessly with Claude Cowork in minutes. Claude is the AI for minds that don't stop at good enough.

15:50Jon Krohn:It's the collaborator that actually understands your entire workflow and thinks with you. Whether you're debugging code at midnight or strategizing your next business move, Claude extends your thinking to tackle the problems that matter. Ah, and you'll appreciate that I can ask Cowork to show me data, such as my sales spreadsheet, and it provides an interactive chart right in the conversation. For problems worth solving, get started with Claude at Claude.ai slash superdata. That's Claude.ai slash superdata. And check out Claude Pro, which includes access to all of the features mentioned in today's episode.

16:19Claude.ai slash superdata.

16:23Jon Krohn:Makes perfect sense. I feel like if I was in your situation, I might be scared because there's reports like, So I did last month, I did a five minute Friday episode, episode 924 on this report out of MIT. You probably know about this, Larissa. It's from the Nanda group that said that 95 % of AI projects fail to show value in production or, you know, never make it to production. So it's only 5 % of AI projects that get into production and then provide business value. So if that was the case with Unframe, of course, your whole business wouldn't be tenable because 19 out of 20 projects, you'd be giving free POCs, the customer doesn't get value.

17:09Jon Krohn:And of course, you know, then you wouldn't have a business either. So somehow you found a way to get a return on investment from the vast majority of the AI projects that Unframe does. How do you do it? But actually, you know, when that report came out for us, it was like, finally, someone is seeing it, you know, because that's what we've been preaching the whole time. And that is what like, it really is the scientific proof of why our model is so superior in this space, to be honest. So think about the other type of AI platforms that you have right now, you have like the prompt to app generator, you know, like I want a tool that does X, it's not enterprise ready, you know, but everyone loves it.

17:50everyone loves playing with it or you have all these um agent builders but then you end up and actually i just had a conversation with um a large insurance company um on the west coast uh just this week he's like well our um business leaders they come to us and they say they want to build agents you know and i was like that is fantastic but for what it's like it doesn't matter like they they we just need to give them something as the it department and they can build their own agents I was like, do you really want to give them something where you don't know what they're going to do? And you're going to have like hundreds or thousands of Asians that are not maintained, that are doing autonomous actions inside your business.

18:28You're highly regulated. He's like, actually, no one has ever pushed back on me like that. I was like, that's why the business impact is actually important for us. And we don't want you to be ending up with all of these resources, like whether that's time or money or both, spent on POCs, but you actually don't know what you're working towards. And so that's actually for us, that report was gold.

18:50Jon Krohn:So yeah, so you've commented on the report, but then how to answer my question, how do you pick projects or how do you pick KPIs, business indicators that you know are going to be a success? How do you get in that 5%, Larissa? Yeah, no, absolutely. So usually when we talk to a customer, they have a few use cases that they have in mind. And what we do is we do a bit of a deep dive on those. So like, what are you trying to achieve? And usually there are some areas either like you're trying to reduce headcount, you're doing, you want to be more efficient, you want to improve the response time to customers to tickets, or you want to reduce costs, you know, and so we really try to drill down with our AI strategists and like, which one is business impact?

19:35Where does the budget, where does the team actually buy into the idea? because you can have as much business impact. If the team that you're building towards is reluctant to change in that regard, is not going to adopt the solution, you're not going to move the needle for your business. So there's about five, six different factors that we look at. We do a half an hour brainstorming session. And usually at the end of it, you have a pretty good idea, just asking the questions and seeing their responses, which one you should start with. And that's usually what we would tackle. Then if possible, we get a few data samples that they share with us.

20:09And if they're receptive to that and the business user actually works with us on getting that POC stood up, then you're on a good path.

20:18Jon Krohn:Cool. Great tips. So when any of our listeners are out there considering building some AI functionality, how do they decide whether to build or buy or do both? What are the kinds of tradeoffs? Interesting one, because also conversations, obviously we see on the news all the time, but also realistically in conversations, especially early ones that we have, we see this a lot. Usually they tell us, well, we're currently in the process of deciding for AI, do we build or do we buy? Realistically, in enterprise level, it's probably going to be both. You can't possibly be building everything. Not possible.

20:56But also maybe you shouldn't be buying everything either. I think most cases like this, probably 80-20 rule, buy the things that you need to get quick impact on your business. Stuff that is simple use cases, maybe. Quicker wins, those things. But what you could consider building, and I totally understand when you want to build that in-house, is anything that has like IP ownership concerns. You want to increase your strategic positioning, things like that. So, I mean, think about your bank, your core banking system. I understand if you want to build that or kind of customer interaction type things.

21:35But if you're talking about like your IT ticketing system or knowledge based search, marketing, content generation, do you really need to build that and have this backlog and like have your internal developers maintain that for the rest of their lives? it's probably not necessary and you would be significantly more cost effective and faster to just buy it in my opinion.