AI agent Fin nearing $100m revenue: Des Traynor, Co-founder at Intercom.

0:00Scaling Europe:Hello and welcome back to the Scaling Europe show presented by Ideal. I'm Seb Johnson. Today I'm joined by Des Traynor, co-founder of Intercom. Des, thank you for joining me today. How are you doing? Good, good. Excited to be here. Thank you for joining. It's great to have you. You've built a phenomenal business. So yeah, I'm really happy to have you. Yeah, thanks for your kind words. Yeah, I'm excited to get into it. And look, you've been building Intercom now for, what, nearly 15 years? Is that right? Yeah, it's been a while. Yeah, yeah, for sure. What are you focused on at the moment? I mean, it's trite to say AI, but my actual job at Intercom is mostly to work with our product design and engineering teams and make sure that we're building all the right stuff according to our customers, but also not just according to our customers.

0:41The big focus for 2026 for us will be, so we're kind of best known as a, formerly we were known as a help desk customer service. Then we went in hard on AI and launched FIN, which is our AI customer service agent. but this is the year where finn kind of broadens its mandate and finn will become a kind of a whole customer agent so it'll handle sales queries it'll handle success it'll handle like all sorts of stuff like that as well so that's the kind of big focus for this year and how's the launch of finn gone so far i mean incredible in so many ways finn is like uh you know obviously i you know intercom is as you would guess with the 15 year comment it's like the majority of my career so i don't have like a load of experience bringing a load of other products to market but finn is like nothing i've ever seen or nothing we've ever seen uh it's you know uh adoption uptake customer satisfaction we have 7 000 customers fin resolves a million over a million conversations a week it's on track to hit 100 million in revenue it's you know it's it's like it's an insanely fast growing insanely popular and valuable product that is basically being ripped out of our hands and like not just by intercom customers but by like zendesk customers and salesforce customers like everyone wants feeling because it kind of plays nice with everyone and how did that go to market work because

1:53Scaling Europe:I imagine that Finn was probably is a very or not an easy business easier sell to existing and to come clients when did you see that transition I guess from it being an add-on to existing clients to actually wow everybody wants this I mean we actually saw from the very start I think honestly if I could do it all different I probably would have launched it everywhere for every community but you have to kind of wind the clock back a little bit like in 2023 like we brought Finn to market March 2023 a lot of the talk back then was about like hallucinations ungrounded inferences can you trust it etc etc so our real kind of focus in in that period was demonstrating this this thing is really reliable and it really works really well and at the time finn's resolution rate was perhaps 25 today it's like 70 69 something in that range so that would mean if you give it a thousand conversation support conversations it'll it'll only hand over 300 of them be humans which is like changes your entire thinking about support but it wasn't like that back then so because of that we thought well you know what this is obviously an essential tool that our customers really love but like i can't you know it wasn't obvious to us at the time that like that everyone else was going to want this very quickly after we launched a lot of the interest we got was from basically because we were out first we were like ahead of everyone else by a huge margin in the market chronologically and so as a result we had a lot of latent demand and it took us probably a few months to realize oh my god we need to build us to sit nicely inside a zendesk messenger or an agent force or it wasn't even called agents force back then probably einstein iq or something like that whatever sales force messenger was called back then but today it does today film works across service now hubspot freshworks you name it it'll play nice everywhere and we're getting a lot of business from that and you know you mentioned you know 60 70 it's solving now how hard do you think it can go i think it'll get to like high 90s um i think they'll like basically if you invert the question what stuff is it not reasonable to expect artificial intelligence to not be able to work out i think you will bump into categories of problems where the actual the right resolution is not defined and what that what that what that i mean like the business has never seen this problem before and literally does not know what to do with it so if you imagine like um let's say um let's just say on your podcast if somebody asked hey we'd like to buy we're like an educational institution but we'd like to buy a sponsorship for two years but we want a discount and we want to have an interstitial halfway through a podcast that's something that you like sebastian have probably never thought about how you would do yeah so it's kind of hard to work out how would finn know what to do in that situation because you actually don't know so you i think that the the residual problems i think will be almost always like uh just never before seen uh things that the business has no idea how they want it to behave And then there'll be a constant kind of little just recurring, like in fast-moving product teams, bugs or UX confusion can kind of happen quite a lot.

4:53And occasionally you'll ship something that doesn't really make sense. And that might mean Finn, if Finn wasn't giving a heads up on it, it might not know what to do. But I think high 90s is a reasonable approximation.

5:02Scaling Europe:And you mentioned that this product is sort of running away from you. I've heard at Product Market Fit be described as instead of pushing the boulder up a hill, you're sort of trying to chase the boulder as it rolls down uh when you're in that situation and the credit's running away from you're getting amazing outbound from customers that you said 7 000 people on boarded how do you decide what to build next that that is the big challenge um this is like you're you're you know i don't i don't know if ai necessarily changes uh this because you're kind of it's a kind of product manager one-on-one here but the way you can think about it is our you know my philosophy is always start with happy customers so are our current customers really happy if the answer that question is no or if you're doing a lot of undifferentiated heavy lifting that we could sort of solve for them in the you know by writing better product or more AI or whatever that's usually the right place to start then if you're trying to expand your addressable market like for example hey let's go big on e-commerce well guess what we're building an e-commerce buying a system inside fit like those types of things are kind of like you know when you build those features it's because you believe hey we're in dealing what we're currently doing and what we want to do is expand the market even further so you're kind of working at what is the actual problem you have if your problem is your customers don't love your product make that your number one focus that's not our number one focus our customers love our product so then you're into like okay well in what direction should we expand the only way in which ai changes this i think is that you should probably dramatically raise your ambition in terms of how many different directions you can pursue just because of cloud code ai engineering all that sort of stuff like it's it's less of a sort of difficult choice than it used to be turns out when when claude's writing one feature all you need to do is that apple tab and jump into a new terminal have another cloud write the next feature so you do have this kind of like um because changing pace of technological development where like saying we're not going to do x now means you'll probably get to it next week whereas it might historically have meant like a q3 or something yeah amazing uh and one of the things interesting about fin is that it's almost being positioned as a you go to the fin website it's like a separate thing right it's not doesn't really get branded under the intercom brand do you see this becoming bigger than intercom yeah um definitely i i think like um if you know what i'm answering there is do i believe that the trajectory of the revenue of the fin business will eclipse the revenue of the intercom help desk business yes um that's like i i think that's you know the ai of version of every sort of like classic sass will be dramatically business sorry dramatically bigger than the classic sass the reason for that is very simple it's doing the work the other product is seats people sit in to do the work over time businesses will want software to do all of the work so they're just going to flock there eventually and we're seeing this big sass sell-off all the public stock markets are way down you've managed to sort of i don't know your guests sort of like accept the innovators dilemma and be willing to cannibalize parts of your own business why do you think you and intercom have been able to do that while so many sales companies haven't been able to successfully build the ai version of their classic sales business there's a lot of reasons uh i'll start at the top um a sort of strong decisive and brave ceo is really necessary here and that's what our ceo on is like it takes it takes a lot of like guts to see something coming that is a really inconvenient truth and embrace it and act accordingly and that looks like you know massive rip-up of roadmaps and reallocation of resources and going all in and buying a new domain and you know you know pushing all of our brands towards that etc like every single one of these decisions is painful and whenever i talk to companies we get reach out all the time hey we need to do what you guys did how do you do it and as soon as you start describing some of the pain that you have to go through some of the cannibalization as you refer to some of the internal strife that you might have to deal with inevitably they say is there not an easier way or like what's the dive version of that like you know like but like the reality is it's not easy if it was easy everyone would be doing it obviously there are other exogenous factors i think like not being a public company probably helps it's a hard message to say we're going to hand back a lot of revenue because we're pursuing this new thing that's as yet unproven which is why i think the public companies have been a lot slower to react to ai because again remember it started off not as good as it is so by the time they got around to spinning up an ai lab they were probably just quite a bit behind but yeah i think like ultimately you have to be willing to make you know you have to first of all really grok the sort of innovator's dilemma itself underscored that like you can't just fight to protect something that is inevitably going to die instead you must make self-harming moves to give yourself a chance of a rebirth in a new sort of product line that is fit to compete in the ai era and everything you have to do there is going to cost you it's going to hurt people internally will not understand it not like it maybe try and delay it or dilute it or postpone it um and then you need to correct it all in the eyes of your customers in the minds of your customers because like in 2023 no one was like hey we want to go all in on ai if we literally just listened listened to our customers i don't think we would have uh done what uh you know done anything near like what we did because customers weren't aware of how good ai was going to be either but now they're all buying ai and thankfully we have it so it's just you know there's a lot to it but i do think the step one is big bold brave decision step two is be willing to make every single painful uh self-harming decision you need to make to transition into being an ai company and how do you balance that i mean it's such an interesting dilemma to be in in terms of like listening to your customers that the classic yc advice is like stay close to your customers listen to them you know do what they say and then there's that other phrase i can't remember who said it but i think it was henry ford who said like if i'd listen to my customers i'd have given them faster horses instead of built my car how do you balance that someone who lives and breathed product how do you know when to listen to your customers versus when to ignore it and kind of make those big bets yeah i mean there's a lot in this and please interrupt me if i go on for too long because like there's so much uh depth to this point so if we just take the henry ford quote for a second they would have said faster horses and like what a proper designer or product here in that uh in that conversation would say is i understand from what you're saying that speed is an important characteristic that you care about in your mode of transport what else bothers you and they might is it oh i hate getting wet when it's you know i can't see at night blah blah blah and you'd be writing in all of these pain points with the current world because it's not your customer's job to solution for you they're not experts in what's technologically possible your customers are experts in what they want and what a good product researcher or pm or designer would have gotten out of that conversation would be a list of requirements that i say speed comfort warmth nighttime variability reliability blah blah blah right they would have taken up maybe cost to run or whatever they would have taken that list down and they would have said no taking uh that's the customer first aspect now we have to look at it from a technological point and bear in mind a technological breakthroughs whether it's mobile or whether it's the combustion engine or whether it's ai you you can't just listen to a customer you have to also listen to technology and say what what are we making possible today that wasn't possible yesterday and in our you know again in in the world of like say um cs right we knew our customers obviously you know they didn't they used macros for a reason because they were like dealing with repetitive questions we like we knew that there was like lots of evidence that suggested they didn't really want to deliver handcrafted tailored artisanal hand typed answers to every support query so like they did clearly that was the thing they cared about yeah but they didn't know what ai was making possible so we had to like basically it's almost like there's a double-sided thing you have to look at whatever is being innovated on the technologically possible and whatever customers want and when you find this bite point where you're like we think we can have a back and forth conversation and we can do some basic retrieval augmented generation to look up the answer to a question when the answer has been previously written in a doc somewhere and we think we can handle informational queries like that and then all of a sudden lo and behold you present that to customers they go well actually shit that that works really well we want that and then you go again and you say okay well what don't they say well we have all these complex queries and all these like you know we have to go and log into systems and do refunds and we say okay and we look and we again go to the ai team and we say things we need to do um follow a kind of a script execute a refund policy see if we can log into third party systems and trigger actions to do refunds they kind of cook it up in the lab and when they're happy they say no we can do this and then we write to customers we say look we can actually do your all your refunds and like that's how you actually work in sort of concert with your customers if we had to just sat on our hands and waited for our customers to be like ding dong can we have some AI, please.

13:52It would have been 2025 and we'd be dead. And you don't have the opportunity to only listen to customers during a tectonic shift in what's technically possible.

14:03Scaling Europe:Yeah, I think you've nailed that, actually. You're right. You listen to the feelings or the emotions of your users and your customers and you've got to be the expert in translating that into the technical... I'm sorry, customers are experts in the problem that they want to solve. Yeah. That's who you're experts in, but if they're really important. Yeah, yeah, yeah. Okay, that's super interesting. Well, I also want to talk about some of the other things you're involved with you know you're like an incredibly prolific angel investor i think you made over 100 angel investments including some amazing companies like stripe in motion how are you sourcing and finding these amazing companies and deciding which ones to back i mean historically i used to they're all like a lot of them were um i stumbled upon them because of intercom if you like right like um they'd either be intercom customers or they reach out because they wanted to like you know but they needed to discount because their or a bill was blowing up or something like that um and then obviously over time you become helpful and useful to certain customers and then like you know you build a kind of a network of people so take like say i'm i invested in a company called real time board which was an online whiteboarding service that grew up to become miro and become this absolute monster business and through that you meet the ceo and the ceo themselves and andre is a great investor and he connects you with a lot of people because of that you know so you kind of you play your network out but i think like the first sort of the origin story of of my actual career writing checks into startups is probably just getting to know startups because intercom is an extremely popular tool for startups and then from there being useful to uh to startups a lot of them thankfully in my case have blown up and then as a result you kind of get exposure to more and more people and then and then so as a result like these days i get probably more far more pitches than i can actually deal with i really can't even entertain a kind of a cold pitch i have to just wait until somebody either an investor or a founder has told me like this thing's pretty good and that's the only time which i can actually justify otherwise i'd otherwise i'd be a full-time investor and i wouldn't be able to work at intercom yeah and are you looking you do you find yourself investing in a certain type of company more than others at the moment are you going big on ai is it the type of ai that fin is or are you going more for infrastructure what does that look like i'm probably going for uh uh like i have this kind of like four requirements for an ai startup which is basically it has to solve a real problem has to have differentiated ai um has to be able to like have pricing power so it's not a thin wrapper or whatever and it has to have like some margin um like some path it's margin let's say so like if there's not deeply differentiated ai in a specific domain um i i kind of check out a little bit and that kind of outrules things like copilots thin wrappers or like things that are like kind of we're selling we're selling tokens you know we're selling a dollar worth of tokens for 50 cents and thinking we're gonna we'll make it up in numbers or whatever yeah i try to avoid a lot of that um so but outside of that yes i'm writing a lot of checks in with a lot of ai companies at the moment but as long as they have to have a really good articulation of the problem it has to be a really good solution to the problem uh and yeah i'm seeing some like really explosive like there's a lot of ai is weird like the amount of times i've had written a check saying like mid 2023 or late 2023 and i haven't heard anything from a company for ages and then i get a mail this literally happened to me like two weeks ago we get a mail i'll be like hey we finally started charging and we're at 17 million in there or what yeah and so like yeah ai is just different that's wild and like i don't i would even say don't care two chickens there because i'm also very very aware of just like think about say coding agents right uh how often the lead has changed hands in coding agents even in intercom we went from like cursor to windsurf to augment back to cursor over a cloud now you know like this and there's probably i'm probably forgetting their codecs and two or three others in the mix here yeah and at any given point any one of these was clearly the future of software development uh so like i just even the businesses that are blowing up i have explosive uh like revenue figures i'm still a little bit like yeah let's see how it actually really plays out like i think we're still in this little bit of an experimental phase of ai where everyone's buying everything for all the things you know prosumer churn is nasty and it'll catch up with you at some stage your growth will dial down and prosumer churn will catch up with you and we'll see how a lot of this shakes out but i still think yeah i mean i i suspect i hope that i have picked a few winners in the ai era because they look to be huge yeah i think that's the thing the market opportunity is massive but you're right i think because uh a lot of there's so much vc cash flushed like flooding the market a lot of these tools are probably cheaper than they will be in a few years which does mean that people are very happy to like buy and try and then drop and move on to other things so you're seeing all of this a lot of the revenue that's been generated by a companies actually coming for other ai companies that have got flush with cash or vcs um but you're right the winners are going to be absolutely huge um like the final talking point i really want to get onto is it's kind of your role your background as a european founder you know europe seems to be having this moment of kind of like autonomy and sovereignty and we're seeing these amazing air companies being built out of europe where we're seeing like real market or category leaders being built here in a way that didn't happen for a long time whether it's the syndesias or the lovables or the 11 labs you know these like true market leaders what's your view overall on the state of european tech how do you think european founders are performing i mean you kind of summarized a lot of my thoughts there like one is like i think in general europe missed the sas um sort of era we didn't really have a lot i mean i don't mean to be dismissive i'll exclude intercom we are a delaware registered company um but i don't think europe produced a lot of like huge sas businesses uh and and then when you look at ai at the application layer, and we should be specific at the application layer in AI, I think many of the winners, many of the, what would you say, the terminal winners, I think Synthesia is going to win their category.

20:02I think Fixer is doing great stuff in email. I think Granola is doing incredible stuff with meeting notes. I think a lot of these companies are just executing extremely well. It doesn't seem to be bothering them or holding them back that they're based in like in either and well a lot of them are in the uk like it doesn't seem to be the same restriction it might have been in the olden days i don't know if that's go to market has changed a bit or i don't know if that's like capital available in europe has changed but it's it's less of a logical oh my god you raise a series a get the hell out of here and go to california like it that's not the path and when you talk to these founders that's not their plan you know they will open sales and marketing offices in san francisco or whatever in new york but like it doesn't seem to be like that the the old recap and like you know um you know basically re-incorporate into your entire business over there it doesn't seem to be compact anymore so in that regard europe's doing great um as to why i don't know like i don't know what triggered it now but um i'm i'm impressed with what's going on in europe and i know i just mentioned a lot of uk startups i should say like there are lots of you know obviously like you mentioned lovable lock on on stockholm lock on on paris etc even dublin we've got we've got a your relatives to our size we're doing well um at the model there we're kind of not yet and uh i mean mr are putting up a really really good offering but i'd love to see like seven more mr al's to actually have conviction that we actually have something in the fight there and uh and then the like the last concern i have is just a like perpetual gray cloud of the eu regulators who just have this extremely weird penchant for publishing pdfs and celebrating them and not realizing that the ambiguity and the uncertainty that they create whenever they say well we reserve the right to put you in jail based of how you interpret our pdf they don't realize the degree to which that holds people back and and i i think they're just so far from technology that they just think that they're like you know it can be totally acceptable to stick a fine on an american company and call that like tech revenue or something like that like i think they need to get extreme like i mean it's not complicated to break all this down but like america is growing faster than europe the majority of the reason is technology the majority of technology at the moment is ai europe is in the middle of regulating ai not actually enhancing or endorsing or being a positive player in ai therefore you can make a reasonable argument that the eu is the biggest obstacle to eu growth and um and that's not good you know so um so i i i do hope that sensibility or i would say returns i'm not sure if it was ever necessarily dare because ai is so new but i do hope that they realize the implications of making it really ambiguous whether or not a product can exist legally in europe it trickles down into like will our software engineers have access to the latest models as customers will you know even things like the meta ray bands can you access ai and them depends right like the amount of stuff we accidentally make illegal that or like or just enough in the gray area yes worth the american company's time so i i do worry about that side of it and i think you know if if they could come up with a cleaner clearer message and a bit more of a kind of air cover for european

23:12Scaling Europe:companies i think it would help a lot amazing well thank you so much for joining i think yeah i think i completely agree that the eu is definitely in the way and with the founders something has changed in the last 10 years uh we are found as a world position to take advantage but we just need to we need to give them the environment to help them thrive uh but it's been great chatting to you guys I've really appreciated it and really enjoyed it so thank you you're welcome so much Sebastian