Albert Invent raises $55M Series B to build chemistry-native AI for industrial R&D

Speaker 1: Here we go.

Speaker 2: What's going on? Nick, how are doing? Welcome to the show.

Speaker 8: I'm doing well, guys. How are you doing, dude?

Speaker 2: We're good. I think we need to warm up this Gong. First Gong hit. We we had another couple other big announcements, but

Speaker 1: We kinda missed the Gong moment. Crazy. It's been a little crazy today.

Speaker 2: Lot of people on. So but this is this is a clean Gong hit territory. So take us through it. What's going on?

Speaker 5: Yeah. So excited to announce today series b 55,000,000 led by TCV. Let's go.

Speaker 1: Great hit, John. Congratulations. Yeah. Great. Well, we'll we'll talk soon. Alright. Yeah. Yeah. Introduce yourself, the company, how you got here, all that good stuff.

Speaker 5: Yeah. So at Albert, we make chemistry native AI. Mhmm. And so we're helping scientists in the lab with the white coats invent faster. Mhmm. A lot of people don't really appreciate the chemical industry, but it's, you know, the entire physical world is made of chemistry. Everything in your guys' studio, everything you're wearing is chemistry. And traditionally, it takes years to invent and launch new products. Mhmm. And so we're replacing paper notebooks, bringing AI into the lab, and help them invent faster.

Speaker 2: And what's your what's your thesis between fine tuning models, training your own models versus using the advancements that are happening in the the foundation model companies and then diffusing that with software integrations, etcetera?

Speaker 3: Yeah. I mean, there's no

Speaker 5: sense in reinventing the wheel if the wheel is working well. But it turns out that in chemistry, it doesn't work too well. And so as part of this announcement, we actually are announcing a new Frontier retrieval model that we'd trained. It's 45 x better than OpenAI and Voyage and the the... Some of the traditional retrieval models. Sure. And what it really comes down to is the models that are trained on the entire Internet are really good at discriminating between this is life sciences, this chemistry, this is legal. But when you get down to what's the difference between a, you know, nanoparticle, encapsulant, in a paint coating and a, you know, some sort of, you know, cellulite there, they cannot differentiate between that. And so we had to build our own models for that, which is which is fun.

Speaker 2: 45 x feels like a different type of lift from when we talk to, like, a a legal company that's like, yeah. We fine tune this, and we do 20% better on this benchmark. Do you expect that to hold for a long time because of the data? Because there there... There's kind of... There's constantly this back and forth, and then the the the specific models get leapfrogged, and then they go back and forth. How are you thinking about this over the long term?

Speaker 5: Yeah. Super good question. So it all comes down to the data and the availability of the data. It turns out that in chemistry, all of the publicly available data is is the the successes. Right? So if you think about science, right, you publish your successes and the failures, you basically discard or you keep in your paper notebook. And so we have customers who are using our platform that have more tokens of chemistry information in their single enterprise than the entire public corpus of chemical information out there. And so that allows us a pretty big advantage to build something that's very domain specific to them.

Speaker 1: Yeah. That makes sense. Yeah. How is how is, people that that work in, you know, chemistry and material science, how are they reacting to model advancements? I can imagine I can imagine, you know, you're working with commercial companies. They see model progress. They probably get excited because they think we can do more work faster. Whereas, like, mathematics that's happening on university campuses, that can be a little bit more jarring because you're sort of like, hey, there was a bunch of work that was being done that that we were enjoying doing, and now there's no point in doing it anymore, and what do we do kind of thing?

Speaker 5: Yeah. So I think it's scary to a lot of the people. Right? I mean, there's... A lot of our customers are 100 year plus old companies. They... You know, change is hard. And the other thing that's different is that you have to get into the physical world to close the loop. Right? With math or coding, right, you can have a model that just reinforce learns on a computer, and you can just set it and run it, and it will just get better and better and learn. In our world, we have to get to the lab. You have to actually run that experiment. You actually have to do that. And so we like to think of, you know, software engineers are turning it to software architects. Right? You're architecting systems. You're architecting, you know, your applications. I think the same is happening now in chemistry, maybe a little bit delayed, where instead of you being the person doing the work in the lab and maybe designing every nuance of the experiment, you're now architecting an experimental, system, and then you're letting the AI go and actually do that work for you. I think, you know, problem solving has become... Is becoming commoditized. So now the question is what is the right problem to solve? And so that is, you know, get close to your customers, understand their needs, and then you can figure out what to actually point the AI at to go solve.

Speaker 2: What is the holy grail in chemistry and material science? Like, everyone in bio says cure cancer, but what is, like, crazy advanced success over the next decade look like?

Speaker 5: Yeah. I don't know if it is one problem, and I think that's kind of the fun of it. Right? I mean, if you want... If you... If Elon wants to go to Mars, he needs better ablative materials for the outside of the rocket. Like, they know how... What the trajectory looks like to send a rocket to Mars. They need better materials. If we want batteries that last longer, we want better solar cells, we want plastics that degrade and don't pollute, we wanna get rid of PFAS that's, you know, leaching into the environment, though, that's all chemistry. Those are all chemistry problems. And so I think there's a lot of holy grails, and our hope is to give the tools to the people that are inventing and trying to solve those problems because if we don't, they're not gonna be solved fast enough.

Speaker 2: Yeah. That makes a lot of sense. Yeah. It's fascinating that there's there's there's so many different opportunities and applications here and probably lots of companies, lots lots of potential customers and lots of work to be done up and down the stack. Yeah. So, yeah, what what an exciting time. Congrats on the fundraise. Very exciting.

Speaker 1: Yeah. Hoping that you're... Like, you know, I've I've gone back and forth on these... On this idea that, when you when you see all these math breakthroughs, you have to assume that like a bunch of other breakthroughs in other fields that are gonna be much more impactful Oh, sure. Directly on people's life like an alternative to PFAS Yeah. Or you know, new new new...

Speaker 2: By Navier Stokes on a daily basis?

Speaker 1: I mean, but I don't

Speaker 2: wanna speak Most people that weren't affected by it. Yeah. That makes sense. Well, congratulations...

Speaker 1: Great to meet. Thank you coming on. And congrats to the team.

Speaker 2: We'll talk to you soon. Goodbye.

Speaker 4: Cheers.

Speaker 2: Let me tell you about Shopify. Shopify is the commerce platform that grows with your business and lets you sell in seconds online, in store, on mobile, on social, on marketplaces, and now with AI agents. And we have Jeff Holden from Atomic, founder and CEO, With us here, Ed and the team, can talk about the