WEBVTT 00:00:00.126 --> 00:00:03.234 Speed, distribution, and teams matter more than ever. 00:00:03.528 --> 00:00:05.502 Ben, thank you so much for being here. 00:00:05.628 --> 00:00:10.332 Where are you finding interesting pockets of innovation in the application layer of AI? 00:00:10.416 --> 00:00:10.962 Absolutely. 00:00:10.962 --> 00:00:15.330 So we've been spending a lot of time in the application and vertical AI part of 00:00:15.330 --> 00:00:17.011 the market at General Atlantic. 00:00:17.263 --> 00:00:18.565 And I'd say two things. 00:00:18.565 --> 00:00:23.185 One, we're seeing that AI labs are increasingly moving into the application layer, 00:00:23.185 --> 00:00:25.537 but they're often focusing on the biggest markets first, 00:00:25.537 --> 00:00:27.553 whether it's health care or financial services. 00:00:27.553 --> 00:00:30.829 And so, are there areas within vertical software 00:00:30.829 --> 00:00:34.904 and vertical AI that you can find effective spots to invest in 00:00:34.904 --> 00:00:36.584 and to build for customer workflows? 00:00:36.584 --> 00:00:40.364 At GA, we've long been believers in the vertical software thesis. 00:00:40.364 --> 00:00:44.186 So building specifically for customer workflows where you really 00:00:44.186 --> 00:00:47.336 understand a market and can build effectively for it. 00:00:47.336 --> 00:00:51.999 And so a couple examples of that are one with Jobber in the field services space 00:00:51.999 --> 00:00:54.519 or with Fresha in the hair salon space. 00:00:54.519 --> 00:00:58.215 And so what that looks like today within vertical AI, let me give you an example, 00:00:58.215 --> 00:01:01.281 is we've been spending time in property management, for example. 00:01:01.281 --> 00:01:03.633 There's a lot of legacy property management systems 00:01:03.633 --> 00:01:07.330 that have built core systems of record where data on the customer, 00:01:07.330 --> 00:01:11.278 on the tenant, and ID and Social Security information live. 00:01:11.278 --> 00:01:14.890 However, building vertical workflows on top of that, 00:01:14.890 --> 00:01:16.402 we increasingly believe are relegating 00:01:16.402 --> 00:01:19.006 those systems of record to dumb databases. 00:01:19.006 --> 00:01:24.509 So what that means is you can build systems of action on top of those systems of record. 00:01:24.509 --> 00:01:27.701 So what this one prop tech business is doing is saying, 00:01:27.701 --> 00:01:32.405 hey, let me start with a conversational chatbot that you can build for tenants 00:01:32.405 --> 00:01:34.337 looking to rent an apartment in New York. 00:01:34.337 --> 00:01:36.563 And so previously when I reached out to an apartment, 00:01:36.563 --> 00:01:42.234 it would take a couple of days to schedule something and go in and tour in person. 00:01:42.402 --> 00:01:46.812 But now all those leasing workflows and back and forth with a leasing agent can 00:01:46.812 --> 00:01:48.492 be automated with agentic AI. 00:01:48.870 --> 00:01:52.440 Secondly, within the maintenance space for a tenant, 00:01:52.440 --> 00:01:54.331 if their refrigerator breaks down and they say, 00:01:54.331 --> 00:01:58.153 hey, let me reach out to a— to my property manager or 00:01:58.153 --> 00:01:59.959 to my leasing agent to get someone to fix it, 00:02:00.169 --> 00:02:03.067 it no longer takes a few days to get a service technician out because an agent 00:02:03.067 --> 00:02:06.175 can go book someone within a few hours to be at your apartment. 00:02:06.175 --> 00:02:09.536 So when you build on top of a system of record and 00:02:09.536 --> 00:02:13.442 you're bringing in all of the conversational workflows and 00:02:13.442 --> 00:02:16.970 conversational data that is going back and forth with the tenant, 00:02:16.970 --> 00:02:20.120 you can build a platform that up-levels 00:02:20.120 --> 00:02:21.926 the system of record in a way that you couldn't before. 00:02:21.926 --> 00:02:26.757 And that there are no longer technology and product moats in the age of AI 00:02:26.757 --> 00:02:30.579 because vibe coding and code-gen makes it so easy to develop new software. 00:02:30.579 --> 00:02:35.199 So what that looks like is speed, distribution, and teams matter more than ever. 00:02:35.199 --> 00:02:38.895 We're seeing wars on AI talent, and it's because teams that can build fast 00:02:38.895 --> 00:02:42.676 and execute for markets are the ones that are winning in the new age of startups. 00:02:42.886 --> 00:02:47.254 Is SMB software still an attractive market in the age of AI, 00:02:47.254 --> 00:02:50.488 or will the next wave of innovation be enterprise led? 00:02:50.782 --> 00:02:52.168 It's a really good question. 00:02:52.294 --> 00:02:54.646 And you know, at Geo we've long been believers that 00:02:54.646 --> 00:02:58.007 SMBs are undertooled and underbuilt for. 00:02:58.007 --> 00:03:02.711 And so we increasingly think that there's a misconception in the market that AI is 00:03:02.711 --> 00:03:06.827 going to be enterprise first because SMBs can vibe code their way to success. 00:03:06.827 --> 00:03:10.439 And so while enterprises do need human-led go-to-market teams and 00:03:10.439 --> 00:03:15.102 compliance-grade products built for them, we do believe that the SMB market is 00:03:15.102 --> 00:03:18.672 really unique because they don't have the time or the resources or sometimes the 00:03:18.672 --> 00:03:21.360 technical know-how to build applications for themselves. 00:03:21.612 --> 00:03:24.846 So we have a business that we're invested in called HighLevel. 00:03:24.846 --> 00:03:31.777 What they do is they build AI agents and AI tools in the marketing and CRM space for SMBs. 00:03:31.777 --> 00:03:35.809 And it's more important than ever now to think about what are the fundamental 00:03:35.809 --> 00:03:40.303 business models and the go-to-market motions that are creating unit economics 00:03:40.303 --> 00:03:42.907 that are sustainable for selling to the SMB. 00:03:43.075 --> 00:03:46.436 SMB has always been a hard place to sell into because churn is high, 00:03:46.436 --> 00:03:48.872 because CAC paybacks can quickly escalate. 00:03:48.872 --> 00:03:51.854 But when you find interesting distribution models with high level, 00:03:51.854 --> 00:03:54.416 it's affiliate- and product-led growth led, 00:03:54.416 --> 00:03:58.868 you can find ways to distribute and deliver to SMBs in a way you couldn't before. 00:03:58.868 --> 00:04:02.228 In a similar fashion, we're seeing AI services rollups really be built 00:04:02.228 --> 00:04:06.975 across the space, whether that's HOAs or call centers or PEO shops. 00:04:06.975 --> 00:04:11.427 And what you're seeing is these previous services businesses were serving small businesses, 00:04:12.267 --> 00:04:14.031 but now they're being rolled up and AI is 00:04:14.031 --> 00:04:15.501 being applied across the stack. 00:04:15.501 --> 00:04:18.189 So whether that's building tools for productivity, 00:04:18.189 --> 00:04:20.794 putting in AI SDRs to really help the growth, 00:04:20.794 --> 00:04:24.238 we're seeing that margins in these AI services businesses are going 00:04:24.238 --> 00:04:28.942 from 10 to 35 percent within a year while increasing top-line growth. 00:04:28.942 --> 00:04:33.688 And so we fundamentally believe that AI is going to be an enabler of SMBs and is 00:04:33.688 --> 00:04:36.293 going to change the market in a way that it never had before. 00:04:36.293 --> 00:04:40.115 And it remains to be seen if consumers ultimately benefit from prices coming down 00:04:40.115 --> 00:04:45.113 across the board or if these services businesses will effectively run 00:04:45.113 --> 00:04:46.457 with higher margins in the future. 00:04:46.457 --> 00:04:51.204 But we're excited about continuing to lean in and build for the SMB. 00:04:51.204 --> 00:04:52.506 Ben, thank you so much. 00:04:52.548 --> 00:04:53.094 Absolutely. 00:04:53.094 --> 00:04:54.060 Thanks for having me.