Toby Mather, Co-Founder of Rig: BREAKING: Rig raised $2.8 Pre-seed

0:00Hello and welcome back to the Scaling Europe show. I'm Seb Jorleson. Today I'm joined by Toby. Toby is a serial entrepreneur who is now building his second venture, which is very exciting. Even more exciting, he's got some great news to share today. Toby, take it away.

0:12Toby Mather:Thanks, Seb. Yeah, today we at RIG are excited to announce our pre-seed fundraise. We've raised$2.8 million from some investors who I've been working with for 10 or 15 years now, including Emerge Capital, who've led the round, and we're in my previous business, Lengumi, Portfolion who joined us, Entrepreneurs First who are my first backers 10 years ago at Lingumi and a few angels including Chris Mayers and Barney Hasio from Clio who are one of our first design partners. Amazing look lots to get into before we get into the round and kind of where you're at and all of that good stuff. Tell me a bit about Rig, what have you built, what's the product, why have you built it?

0:49Toby Mather:Absolutely yeah Rig is building the underlying tools and building blocks that let people use all their internal data with AI. So that's basically infrastructure, building blocks to ingest data from all of your SaaS tools and data sources, connect it all up in a warehouse, build a context layer on top of your existing warehouse or your new warehouse, and then connect all that data up so you can build custom data apps, automations, agents, and so on, without having to wrangle 15 APIs or 15 MCPs into a clawed skill. So it's this headless system that Teams now using to build with internal data. And what kind of things are people building?

1:28I know that you mentioned Barney from Clio's coming on as an angel. You mentioned that he's a design partner. Can you talk about kind of maybe some real life use cases to bring it to life? Definitely.

1:38Toby Mather:Yeah. So when we started working with Clio, we were trying to solve this challenge of teams across the business in this very fast scaling environment, and needing to work with data to solve very specific problems. And the analysts were the bottlenecked that because they had all this context about the data in their head. And so the first case we were tackling was as a dispute team, say in FinTech, how can we help you put in a user ID somewhere in Claude or an internal app, and it can pull together the user transcripts and all the payment activity of that user and all the messages that users have with the chatbot to form a picture of whether we trust this user or is this a legitimate dispute claim or not.

2:18Toby Mather:And that's sort of one example of this type of heterogeneous data-heavy workflows that teams are now trying to solve for with AI. And you can do a lot of the orchestration of the AI bit with Claude, but figuring out which data you've got to pull, where does that live in your system, how does it join to all your other data, that's this gnarly context layer piece that Riggs takes care of for Dio. And so it kind of ingests all the data, surfaces it, and then just enables people to build the applications on top. That's right, yeah. Sometimes we ingest data. So for a startup, we can act as their full data stack already.

2:52Toby Mather:In Clio's case, they already have a big data warehouse. They've got, you know, hundreds of thousands of models already. So we sit on top of that as this kind of data brain, this context layer. And then they can sit in Claude or Cursor, their favorite tools, pull in the data they need just by chatting to it. And then on top of that, orchestrate automations into Slack or automations into Notion or some kind of data app which they can host on Rig. And we found with some of our early clients, one of the problems that sort of we solve unexpectedly well is people sometimes reduce amazing things in Claude and then start sending HTML files around to their colleagues.

3:26Toby Mather:Be like, look at this. And that's not very secure. You can't see what the underlying data execution was behind that. And so one thing you can do with Rig is much more happy in Claude. You can say, great, save that to Rig, please. And it puts it up in this governed, shareable data application that your colleagues can now access and use. Interesting. That's actually one of the most painful points I always have of like exporting things and sharing it with my team and for me it's always like graphics or images but it's always like claude is so good inside claude it is actually taking it out and connect it to other bits that i'm still struggling with um but why this business you know why was the you know you you are building again you you built lingumi your previous business for like 10 years exited and then you kind of stayed on board for a couple of years after that why then is this the topic or this the issue that you felt like you know was pulling you back into the arena Yeah, it's a great question.

4:15Toby Mather:And I suppose preschool education to B2B data infrastructure was not an obvious pivot. But building a scaled up B2C or direct consumer brand, you're dealing with data every single day. And I was the data team at Lingumi, and I was up to obsessed with making sure our data infrastructure worked well to help us make good decisions as we scaled from nothing to 2 million users. And because especially we're doing business in China, we still had to build our own infrastructure to do that. And that became this unexpected competitive advantage against the market. So we can make decisions on how users were behaving, build the timelines of how a user would behave through the product and so on by taking advantage of all our, say, time series data.

4:53Toby Mather:And this was all pre-LLMs. When I then landed at Novakid, where I was leading the product to the data function after they acquired Lingumi, I hit this really basic problem, which was customers were writing to customer support every day, asking how their child's lesson had gone. And customer support would have these generic pre-canned answers that would be sent out. And so we connected up an 11labs voice agent on the parent portal, and the parents could call up and talk in Turkish or Polish or Hebrew about their child's lesson. And the 11labs agent, I thought, this is great. We can now get it to tell the parent how it went.

5:27Toby Mather:So I approached our data team and said, how can we get the data into the agent? And this turned out to be a surprisingly difficult problem. Because it's not deterministic paths of data, which is called reverse ETL in the data world. You're not just sending the same data out every day. You need the agent to dynamically and securely access the relevant data for that user on that day and understand what data to fetch. And I thought, this is a huge and interesting problem that's emerging in the AI world. And so solving problems like that, and for me, rig runs on rig. One of the problems I had was after I've signed a contract with a client, I want to generate invoices for that client.

6:07Toby Mather:And that is another data-heavy problem. I need to pull in the context of the contract. I need to prepare an invoice from it, figure out the billing contact. So yesterday, in about an hour, I built a data app on Rig that generates all my invoices for me. But these problems, they're very heterogeneous. It's a horizontal platform we're building. but they all root back to this question which is where and how can i connect up all these different bits of internal data into a clear picture that an ai talk can act upon and do good work with amazing and uh as you mentioned you've just raised this great round what's what was compelling i guess like what kind of gave the the investors the conviction i know they've worked with you for a long time and they obviously know a lot about you was there anything else where they're like this is What gave them the real conviction to, I guess, double down on you as an individual?

6:56Toby Mather:The thing that seems to be challenging for investors today is that AI's moving so fast that they're not sure if the idea they're investing in is going to survive the next model release. And so we went with investors who knew me or knew my co-founder very well, who we worked with for years, and who know what absolute animals we are. We're just complete machines. And so that we would take advantage of whatever's changing in the market and adapt to it rather than sort of being crushed by it. So, you know, I've worked with the team at Emerge, in fact, even worked for the team at Emerge at one point a very long time ago, and they've seen me operate, and they've seen me operate through very challenging circumstances in Chinese education regulation.

7:34Toby Mather:So I think they took a head of feeling that maybe we could tackle some of the challenges emerging in AI. Oh, that's really cool. I guess, yes, I guess when you're this early stage and when things move so quickly, it is, yeah, I haven't heard that before, that it's actually like, they have no idea you have no idea no one has any idea if ideas will stand up in six months 12 months it's about adaptability and yeah and being able to and yeah investors will always say you know we only at early stage back the team but yeah you know my empirical experience is that tends not to be true and they'll say it but they have a lot of them say oh well you know we worry that you know context layers is a busy space or something like this and i think what they're saying behind that is we don't believe you can win in this space which you know fair enough and that's why working with investors who've seen you operate over a long time especially if you're a second time or third time founder um is it is it is a good way to raise around they trust you they back you and they're going to go the the distance and what are you going to be doing differently this time like what lessons i guess i guess two questions actually the first one is going to be like lessons like what have you learned for building the first time that you're now applying the second time and then i'd like to ask about the environment being so different this time versus 10 years ago.

8:43So let's start with the first one. How have you changed your approach as a founder second time around?

8:50Toby Mather:What I learned in the first founding experience is that as soon as you get any scale, most of the problems become about managing that scale. And so if your product market fit is not really locked in when you begin scaling, then you never have the chance to go back and rebuild the foundation because every day your customers have questions and problems you have to solve. And so with RIG, we're keeping the team really small and we're being incredibly capital efficient at the beginning while we just drive product market fit and really listen to these first customers and solve their problems before we start to scale it up too much.

9:22Toby Mather:And in B2C, you can imagine that problem is intensified by the scale, the natural scale of having immediately hundreds or even thousands of customers as we did. With RIG, we're working with more than 10 customers already, but in scale terms, that's very small compared to B2C. So we are just in Slack every day focusing on their problems, solving their problems, and thinking about how we scale those across customers before we invest heavily in a sales motion, for example. Interesting. And then I imagine building today versus 12 years ago, 10 years ago is very different. The tools available, the speed of which things are moving.

9:56How has that changed the way that you are approaching building this business from day one?

10:00Toby Mather:The thing that's changed building businesses in the AI era, building an Opus baby, as I call it, so a startup born after the release of Opus 4.5, is there's two things. One is about how the founders operate, and one is about how you hire and build your team. As founders in this Opus era, we can build and ship product. We can talk to customers, and we can market and sell every single day, all at the same time. So we can parallelize work. And so the core criteria, I think for a really great founder today is high endurance, high context switching. So I can run 10 ages all day and have conversations and sell.

10:37Toby Mather:I think that's a good quality in a founder. And my co-founder is the same. When we think about talent, we apply that same question. But we now say, instead of requiring decades of experience, we're looking for somebody with a really, really steep curve, because you can use Claude or whatever your AI tool of choice is to solve problems with you and for you. So we are hiring instead of 30 something year olds who've been in their career for 10 years, hiring 21, 22, 23 year olds sometimes who have no experience, or they might be older, but be switching into startups. But looking for people who have this immense appetite to learn, who are already living inside AI tools day to day.

11:18Toby Mather:And we just put them in our repose and in Claude on their first day at work. And we say, you own these 12 things, figure it out from here go and we'll just teach you when you get stuck or help you when you really mess something up like you know do a bad database migration that's so that's so interesting can we talk a bit about what that looks like in practice so you are hiring people with next to know experience and you are indexing i guess not even necessarily on experience within ai tools themselves but it's more about a propensity to learn and to have agency to take action how do you hire for them you know what are you looking for in those either on a cv or an interview to make sure that your you know traditional cvs are all around skill set and experience and you know what have you learned in your career today this is like actually i don't care anything about your career i really only care about your propensity to move quickly to learn to take agency how do you how do you look for that the things that make a great ai native employee are not that dissimilar to things that made a great employee pre-AI at the propensity level.

12:25Toby Mather:They just need less experience now to build the same skill base. So at a propensity level, I'm looking for their capacity to context switch, their capacity to learn fast, and their endurance. So for endurance, sometimes you get easy shortcuts, like they're already an athlete or something. And I think, you know, having ranked highly in a sport or being a long distance runner or something probably is a good mental shortcut. If you can't see that, you might be looking for someone who is a really experienced video gamer, for example. But someone who's shown they can persist at something for many hours in a row probably is indicative that they can manage five or ten terminal windows running long complex tasks with debugging for 10, 11 hours in a session.

13:06Toby Mather:And then the ability to learn. To be honest, I think if they are not already experts using Claude or something similar to that, they're probably not that good a learner in an AI context because if you're not fascinated by, obsessed with and using these things every day you're not going to be a great employee for a startup that wants you to be doing that. So we certainly look at what have you built and how are you using Clawed in your day-to-day life that kind of thing in the interview process. And what do you think the implications are for this in careers more generally? Because traditional career paths have always been around actually building an experience, climbing the ladder you start as an associate, become a head of or a VP, especially in the startup world.

13:45How do you think this impacts general career paths in terms of specializations versus general career paths? And then also seniority. Can you see these very junior people that you've hired becoming a head of go-to-market for you in a year's time or head of marketing or head of operations? What does that career path do you think look like for them? And then what are the implications for the wider world of careers? is?

14:09Toby Mather:I think the AI native career path is compressing. You can become senior very fast. Your ability to become senior now is not going to be about your skill level. You can reach a very high skill level very quickly now. It's going to be about your decision-making capabilities. Can you make two-way door decisions very quickly? Can you make good decisions more often than you make bad ones? And also your interpersonal skills. Can you sit with the clients and sell and convince? can you motivate a team of people to learn something new or do something new or turn on a dime to go in a new direction as AI changes around you?

14:43Toby Mather:And when I look at our younger couple of team members, those are still the people skills and the sales skills and the client-facing skills they need to develop. But the technical skill set is growing even faster than mine at the moment, I would say. That's so interesting. And how do you think the team structure will change over time because at the moment, I guess, you're a very small team. You've got junior people who are essentially like helping to manage across the business entirely. When you look at like a year or two year, when you get to 10, 20, 30 people, what does that look like? Does it look like functions in the old sense where you have a dedicated team on each kind of vertical?

15:19Does it look like almost like an army of more junior people who are running across the business, constantly solving problems, iterating? How are you envisaging that down the line?

15:28Toby Mather:One of the challenges with functional teams historically has been, in a B2B SaaS context, if you're talking to a customer success manager, when you express frustrations with the product, you just have to hope that it's going to filter back to product and product's going to filter it down to engineering. And maybe a year later, you get the thing you needed. Today, my customer success manager is also my product person and my engineer. when a customer says, you know, yesterday one of our customers said, we want to be able to send a GitHub Action webhook to update our context layer from a release.

16:01Toby Mather:And by 2 a.m. the next morning, that was live. And so he woke up at 9 a.m. and he got to work and his feature was ready. So I suppose what I want to do with my team is build a team that are, everyone and the infrastructure that we build at RIG supports this, everyone is able to talk to a customer, whether that's a sale, a POC customer, or a landed customer. Listen to them, find their frustrations, be so expert in our products that you can guide them to the right feature if it exists, but also be able to take the request, work on that request, and ship that if it's a good feature that's scalable across clients that same day.

16:37Toby Mather:And it sounds really radical, but if we just work through the bottlenecks, do you have continuous integration? Do you have continuous delivery to production? Well, yes, everyone's aspiring towards that nowadays. Okay, great. Are your team experts in your product? If they're not, then why not? There's no excuse not to be an expert in the product. And then lastly, are they good at communicating, talking to humans, working out what's parallelizable? That's probably the hardest skill now. That's a thorny human communication piece. So that's what we'll focus on. And tell me about vibe coding. I've read some controversial reviews that you've had on vibe coding, which I guess is sort of like, maybe counter to the narrative that's kind of persisting at the moment.

17:12But yeah, tell me about it.

17:13Toby Mather:I think people should vibe code as little as they can on anything that's not core to their product or what makes their beer taste good, as Bezos once put it. People at the moment, because core can write code so fast, are writing lots of code, say building internal vibe tools. And the thing people always underestimate is what's called TCO, total cost of ownership. The cost of maintaining and looking after security and scalability and dependency management and upgrades and all of that stuff on the tools you build is very large. So in RIG, we have 29 SaaS vendors already eight months in. And now we are just Vibe coding one internal tool, our first one.

17:54Toby Mather:That's not our core product and engineering process because we were so upset with our vendor. We could not find a better vendor. So we thought we just kind of have to solve this problem ourselves. And that's to build our sales room, our sales pods, because we wanted to integrate all the internal data we have, client calls, usage data, et cetera, up into that pod. And we couldn't find a good solution to do that. So, you know, my take on VibeCoding is AI engineering your core product so you can ship faster for customers should be 100 % of your energy. And then if you cannot find a specialized vendor who can solve all the problems behind that, then maybe you build an internal tool.

18:30That's so interesting. And that's the one spot that you haven't been able to do. It's like a, yeah, sort of sales. because at the moment I'm vibe coding almost my entire admin because I think my business is very unusual in the sense that it's like it's social media and it's relationships, it's contacts and like a CRM, it's like there's not really a CRM, you know, like a very non-traditional CRM in many respects for like a content creator compared to a proper B2B size company.

19:00Toby Mather:I think on that, the future of vibe coding is there'll be thin UIs on top of standard primitives. So for you, I can imagine your front end for your data store will be a bit different to a typical CRM, and I think you should vibe code that. But I don't encourage people to be vibe coding their database, their data warehouse, their piping of data systems together because there are good market primitives that solve these problems. So having Superbase as your database is a good decision. Custom UI on top of that might be your choice to do that. Yeah, that's interesting. Okay, well, look, we're coming out of time.

19:32You've just raised this round. What's top of priority? What do you want to achieve in the next 12 months? What's going to get you to that next round, that next milestone of success?

19:41Toby Mather:My goal after raising money is to spend as little of it as possible until the product, go-to-market, sales, onboarding process are so repeatable that we feel we can now start spending this money towards growth and that growth take us to the next round. So today, you know, we've grown revenue 100 % since last month. And I think net we've spent about$500 or something like that. Amazing. I'd like that to be$0. So our goal is keep scaling this early growth. But that's not because we want to get to the next round. It's because we want to show that our go-to-market and our product process is completely repeatable before we throw some fuel on the fire and start bringing in account executives and scaling up a go-to-market motion.

20:21And who are you targeting? I imagine you're trying to get a lot of customers from different industries, different verticals to try and sense the products, find real momentum? Who are those early people who you're trying to hit who should be signing up, using the product, testing it out?

20:36Toby Mather:The hardest thing about building horizontal software, like if you look at Zapier or Retool or Airtable in the previous generation, is there's not a single ICP. We're not saying we're a tool for CFOs. The people we find tend to buy and bring in TestRig are a really AI-leaning person. That might be a really AI-forward CEO or an AI-forward CRO. who wants to build specialized data-heavy workflows or apps internally. So at Clio, we see a lot of the risk and disputes and so on work happening, but people across the business are using it, hundreds of people at Clio are using it. At BirdieCare and other of our clients, we're brought in by the revenue operations team to solve some specific work around data inside RevOps.

21:19Toby Mather:But actually, the finance team are now using it to reconcile CRM book deals to final invoiced amounts to figure out the comp plan for for ease. So it's not a specific buyer, tends to be a non-technical person who's got really into AI and wants to rip out a lot of the legacy stuff in the organization, build a strong data foundation, and then start automating stuff on top of that. Amazing. Well, if you're listening, you're interested, get in touch. But Toby, thank you so much for joining me. Thank you for sharing your story. Congrats on the news. It's very exciting. Thanks, Seb.