Attio raises $52M Series B led by Google Ventures to build an AI-native CRM challenging Salesforce

right up the charts. Yeah. Send it up to the top of the charts. Uh, our next guest is in the reream waiting room. We have Nick from Adio coming into the studio. Sharp. What do we got? Fantastic name. The news. Give us the news. What's happening? Welcome to the show. Hey guys, how you doing? Great.

Uh, late for you, right? It is. It is. Uh, especially when you are you've got two toddlers asleep in the house. This is um well past bedtime. Yeah. Yeah. Well, I know how that goes and uh I'm Yeah, I'm glad you were able to Always good to get them.

It's good to have like a meeting set for, you know, somewhere around bedtime. It's like extra incentive to have bedtime. Yeah. And and also dodge the duties of putting them to bed. So, it's all good. Thank you. Good. Good. Uh well, big day. Big day. Uh congrats on on the raise.

I have the uh the gong mallet here ready if you can give us uh the details. Absolutely. So, we are very very excited to announce our $52 million series B um led by Google Ventures with lesson. There we go. Sorry to interrupt. Um fantastic. Um breakdown uh give us I mean first time on the show.

give us kind of the full kind of abbreviated but full history of of the company uh and how you guys got to this point because it hasn't um uh is a big milestone but uh certainly been at it for a little while now. Yeah, absolutely. No, big long journey.

Um so uh you know fundamentally what what we do is we build an AI native CRM and uh what that means is a very very flexible data model uh allows businesses to mold AIO to to their go to market motion rather than having to do it the other way around which has been the way software has worked for a while. Yep.

And then uh a massive amount of data ingestion which gives us all of the context to automate action essentially. Um, CRM is a big product to build, a very big product to build. And there's not really an MVP of a CRM. Uh, it's a it's a big ask for a customer to adopt it.

And, you know, we we've got about 5,000 companies that that that have now adopted AIO. But to get there, there's a lot of engineering. And, uh, so we did three years of of engineering in a, you know, in the sort of startup desert pre-launch. Um, pre-launch.

Uh and uh yeah, so we did three years pre-launch of of just hard engineering grind and uh launched about two years ago. Incredible.

Um what uh yeah I guess like maybe talk about kind of the inflection points at what stages kind of uh kind of where LLM started to really unlock value for for customers and where kind of where you had those initial sparks because I think like when people for better or worse every CRM provider now says that they're AI native.

So I also want to understand like what what does that actually um what does that actually mean in practice and what kind of advantages do you have as a new player um that that hasn't you know been in the industry for a couple decades at this point?

Yeah I think I mean I think fundamentally we have two really big advantages. So one is the quantity and resolution of the data that we ingest and building a data model that can do that. So LLMs work great when they have lots of context. Um and uh you know you need to you need to feed them that context.

And so one of the things that uh you know we we we worked one of the first things we worked on was this data model and being able to instantly ingest a company's uh you know emails, calendar, product data, etc. etc.

And so that was the bit that really took you know took took a long time to build especially when you need to do it at the scale you have to do it at for for CRM.

The other thing that um is is you know a big difference between um aio and and sort of legacy products is that we think that or we have a thesis that code is becoming the new no code.

So the idea of of no code and these kind of very complex UI uh that you have to learn and very steep learning curves you're not learning how to code but you're learning something pretty complicated um and uh you know especially when you start to hit up against edge cases etc then the UI become extremely complicated and so one of the things that we've sort of engineered into from the offset was the ability to execute code inside of your CRM and you think if you think about what has made Salesforce successful is this infinite flexibility uh through very very complex um engineering processes with a you know with Apex and a language not many people know and huge amounts of difficulty.

But what we what we do with that is build in this sort of native code execution very early and uh LLMs are are brilliant at generating code especially in these kind of safe um somewhat constrained environments.

Um and so that's another thing that we think is going to especially you know as this plays out is going to be a really really big deal. This idea that software becomes so much more malleable because code is uh so much easier to generate.

How do you think about the right like where where's your focus on on the customer side? Uh again CRM has so many different applications within different companies. uh and it's in in the long run you can serve everyone but what what's the focus been to date in in the in the kind of 5,000 you mentioned? Yeah.

So the focus really is uh is um is on uh you know what we would consider builders. So so people who like to get stuck into building things um the rise of the GTM the go to market builder is is pretty is pretty apparent now. Yeah. Um and uh so about 50% of our customers are um are tech companies.

Um and and and even the 50% that aren't uh tend to be people who really want to get stuck into building something. Um but but you know as as you said CRM has a huge range of users. Uh and we think deeply about the kind of you know the consumer of the CRM that's the AE the SDR etc.

um all the way through to someone in RevOps um or or you know building out really complex um go to market motions and so we have to serve all of those customers um and you know we we focus a lot on making the consumption of the CRM super seamless as well and obviously AI has been incredible for that you know massive reduction in data entry and and stuff like that.

talk to us about uh gross margins. They're they're in the news uh with a lot of companies sort of like just reselling inference. Obviously, your business is very different than that.

It doesn't feel like I'm just paying you for tokens or or just output, but uh are there any examples of like customers where you've looked at their usage and you've been like, "Wow, this person somehow triggered like a million a million different calls and and ran up a big bell.

" And uh are there any are there any lessons that you've learned from just like building an AI product that is on the back of inference and does have real costs that are somewhat variable? Yeah. Yeah. Absolutely.

And you know the uh like all good startups the um the the sort of approach we took to this initially was um it's not our focus right now.

in the last sort of six months, 12 to six months, um we've actually made incredible uh you know in incredible um gains there and we're in a very very good place now and that has been a combination of um I mean that's that's a combination of models getting uh more cost effective and us learning to use them better.

The one of the things that we can do for example is we can uh we we can decide at runtime which model is best suited for a task. So I think this is now becoming a relatively mainstream strategy. But if we're digesting an entire call transcript, we don't need to use the best reasoning model.

Um we can do sort of compaction first on a faster, cheaper model um and then put less tokens into a better reasoning model. Y it exactly. Yeah. And so it's been a big engineering effort, but with a combination of all of these techniques um you know we have very very good uh margins. Yeah.

The other thing is um a lot of the the the the sort of the value that the LLM provides is is very apparent to the end user. So it's work done uh and that's changing the way that we think about pricing as well.

And so you know CRM has historically been just a seat based uh product but I think increasingly we'll see that that will that will change.

Um and you know we now have a hybrid uh pricing model which is there is there is a seatbased aspect um and we give sort of very generous automation uh and uh sort of AI credits to those seats but if you want to go really wild and you want to replace a lots of work then uh you can buy credit to do that. Yep.

That makes a ton of sense. Well congratulations in the round. Thanks so much for hopping on the show. We will talk to you soon. Have a great rest of your rest of your evening. Absolutely. Thank you guys. Yeah. Nice to see you. See you. Up next, we have Sean from Coinbase joining. Uh, I want to get his reaction