AI Law Firm Hits $100M But…

0:00Do you remember that AI law firm, Harvey? Uh-huh. They raised a lot of money. Do you want to check how much they raised real quick? I'm going to tell you something. So Harvey just hit$100 million ARR,$8 billion valuation, 50 of the top MLAW 100 as customers. I guess that's the top law firms. Their last round, October 29, 2020,$150 million at an$8 billion valuation. Okay. So Akash, Gupta, and then who backed them, by the way? I don't know. And before that, they raised$300 million at an$8 billion valuation. I think a lot of the who's who is investors in it. Sequoia led the$300 million round. Andreessen, Kleiner Perkins.

0:44Probably in it. Yeah, I can't see all of them. I see it. KOTU's in there too. Yeah, Andreessen's in there. Kleiner Perkins, Sequoia, KOTU, and EQT. Elad Gill, he's in a lot of popular companies as well. Yeah. Cool. So the reason I'm calling this out, Akash Gupta tweeted this out. So again,$100 million in ARR,$8 billion valuation, 50 to top MLaw, 100 as customers. But here's a problem. So you know Microsoft Copilot. When you look at usage for Copilot, you look at Office 365. So 400 million users have paid for Copilot subscriptions two years after launch. Less than 2 % of people of those users actually use it.

1:19And this is the same thing that you're kind of seeing with Harvey too. People aren't really using it. They're just paying for it. And I think the point here is how is Harvey going to grow into it? Because you have a lot of people that are just paying for this stuff and not using it. And I remember maybe two weeks ago or so, someone was talking about how they bought these, like, it was kind of satire. It was like, I'm a CMO and, you know, I bought 2 ,000 Microsoft Teams subscriptions or like AI subscriptions to make it seem like I'm doing something with AI. And then when someone asked me, I just said, hey, progress is moving up into the right.

1:50We're looking at usage, right? And so it's a lot of, I almost want to say it's almost theater. It's almost AI theater. That's what I would say it is. to make people feel like they're doing something with it. Yeah, and I'm not saying Harvey's a bad company. It's probably a great company, but you're not worth$8 billion when you do$100 million in revenue, even if you're doubling every year. It's just too much. Even if you're tripling every year, your valuation is just astronomically high because$100 million in revenue does not mean$100 million in profit. And I was watching Orlando Brava, I believe his last name is, from Tomo Brava, the big private equity fund that gobbles up most SaaS companies.

2:28He's like, look, the problem with AI is the valuations. It's not that these companies are bad. It's like, if you have a$10 billion valuation and you have$100 million or$50 million in revenue, he's like, you know how much more revenue you have to add to get to that$10 billion number? He's like, it's just really unrealistic. And he's like, people are just overpaying for this stuff. And I think it works for venture capital. I'm not trying to bet against the Sequoias or the Kotus because they can make a lot of bets in their fund and they can have one or two winners and that's all they need to make a crap load of money.

3:00It's just like you invest in one Google, one OpenAI or one whatever it may be, right? WhatsApp was there, right? And it's like, they're fund makers. Not only do they make funds, they return so much, everyone becomes rich off of it. And then the other deals, whether they win or lose, it doesn't matter. And that's a model of venture. But the Tom Abravas of the world play a different game in which they need most of their deals to work out because they're buying based off of revenue and profit and they're allocating large amounts of money and they're doing later stage. I'm going to read some of the numbers here while you cough.

3:34So I was coughing right before. So if we look at trajectory for these guys over here, okay, so$5 million seed round in November, 2022, and then it rolled it out to 3 ,500 lawyers by early 2023, then$21 million Series A, then$80 million Series B, then$300 million Series D at a$3 billion valuation. That was this year in February. Then the$300 million Series E at$5 billion four months later, and then$160 million more at$8 billion in December. So$760 million raised in 2025 alone. Now, a former Harvey employee claims that retention hovered around 35%. And then you have on our Reddit, r slash legal tech, practitioners report that usage concentrates among junior lawyers and return use is low.

4:20A Harvard Business School case study from March 2025 explicitly states that while Harvey had successfully focused on aggressive customer acquisition, retention was now the key challenge. And so all that to say is I hope it works out for them. I think they're smart people. They've got the runway to figure it out. But I feel like we're seeing more and more of this happening, whether it's like Copilot or Harvey or something like that. You want to make sure whatever it is that you're building with AI, someone earlier asked us what we're doing with AI. You want to make sure it's sticky and people are coming back to use it.

4:47That's the name of the game. It's not monthly recurring revenue. It's almost monthly recurring value. Yeah. What do you mean by Copilot? What were the stats again from Copilot? Because it's included in your Office subscription, right? Or your Microsoft subscription. Yeah. So what I'm seeing here is this. Microsoft cut Copilot sales targets by 50 % last month after enterprise customers refused to move beyond pilots. Less than 2 % of Office 365's 400 million users have paid Copilot subscriptions two years after launch. So we've tried company-wide a few of these different versions, whether it's GPT or you want to use Gemini for your corporation or CoPilot.

5:23We find a lot of people use them, but not as much as one may think. And I've even went in there with a few of my people. I'm like, why aren't you using it heavily? Let me see what you do on a daily basis. And I told them in advance, this won't affect how I judge you or anything like that. I'm just trying to understand how people are using it so that way I can get better penetration throughout the organization. And what I found is when they showed me how they used it, it wasn't bad. It was just, it would create a lot of slop for them. And then they're just like, man, the amount of time it takes to fix it.

5:57It doesn't mean that it won't eventually penetrate into the corporate world. I just believe the technology isn't there yet where a lot of people can rely on it for majority of their tasks, at least in marketing, right? I'm not saying it won't get there, but that's what it is as of now. Yep. So anyway, I mean, I think we're going to see a lot more of this in the interim. I do fully, and I think you believe this too, that it will get there. I just think sometimes you don't want to put the cart before the horse. And don't get me wrong, I'm the first one to be extremely excited and optimistic about AI.

6:28Neo's a little more pessimistic when it comes to new things, but it's good. It's good to have a balance. So speaking of AI, I read this tweet over here from Jason Lenkin. So basically, AI-infused teams are three times more productive. So Jason Lemkin, okay, five things every CRO, CMO should know. He did, I guess, so top five. AI agents are better than middle-of-the-pack SDRs now. Not the best, but better than average. That's enough to change everything. Number two, deploying the first agent is your job. If you haven't trained one yourself, you're becoming obsolete. Number three, pick two vendors, not 10.

6:59Each agent takes 30 days to train. Bake-offs of eight to 10 are impossible. Number four, don't sign until you've talked to your forward deployed engineer. Deployment support greater than features. What a forward deployed engineer means is they just have an engineer that's helping you with setup and onboarding. And number five is AI infused teams are now three times more productive. Kyle's team at Owner, I think it's Owner.com is proof. $3 million quotas now. And so I'm curious from your side right now, what are you testing with new sales stuff? So I'll give you an example. I'll start from our end.

7:30So we use like a blue texting link. You know how when you get a text on iPhone, it's blue, right? So we use a tool called LINQ, L-I-N-Q. And our response rates on that are 70%, I think, if not even higher. So imagine you have a lead come in and they get a text to their phone and it looks like an iPhone. We have people, prospects that actually thank us for texting them because they believe it's real. So that's one example. And then we're also looking at what other, we built some agentic workflows internally we used to quickly qualify people and move them along. and even visitors to, I think like Salesforce just acquired a qualified GTM or something like that, like earlier this week.

8:11And that's one of those agents that will scan your entire website, look for ICP fit and try to nurture people, even though they haven't raised their hand. So what are you guys doing around that right now? Sure. So a lot of the stuff you're talking about, like sending text messages, emails, we were already doing beforehand. What AI has done is just helped make the content better and more personalized at a faster rate. I went over some of these numbers with one of our divisions. And one of our divisions in the United States saw around a 13 % to 14 % increase in sales in revenue per rep from AI. We haven't seen what, or maybe I'm misunderstanding what Jason Lemkin, I believe his last name is, is saying, in which people are getting 3x more efficient.

8:55We are seeing more efficiencies. We're seeing more revenue. He didn't claim that people are seeing 3x more dollar value. It just sounds like people are able to do more. We haven't seen like 20, 30, 40 % lifts in revenue per sales rep because of AI yet, at least. So let me, if you've ever built a website, you know how tough it can be to keep a strong design while ensuring site performance is fast. That's where Framer comes in and it totally changes the game. Framer is the design first no code website builder that lets anyone ship a production ready site in minutes. I recently built a custom landing page in just a few hours.

9:31Animations, fast load times, responsive layouts, all without writing a single line of code. It was easy to use and the end result was polished enough to look like a developer spent days on it. One of my favorite things is that Framers AI handles translations with a single click. So your site can be up and running in multiple languages almost instantly. And with real-time collaboration, your whole team can jump in and work together. No issues with versions or miscommunication, whether you're starting from scratch or using one of their templates, Framer lets you focus on design while handling the technical side for you.

10:01Animations, responsive layouts, search optimization, and all the things that matter. Ready to build a site that looks hand-coded without hiring a developer? Launch your site for free at framer.com and use code MS to get your first month of pro on the house. That's framer.com, promo code MS, framer.com, promo code MS. Rules and restrictions may apply. You've been talking about launching a business for years. You've got the skills, the ideas, the edge, but no product sells itself if it doesn't exist. 2026 is your reset button. It's time to stop thinking about it and actually build it. And there's one move that sets everything in motion.

10:36Shopify gives you everything you need to sell online and in person. You get hundreds of beautifully designed templates, all customizable. Their AI tools write your product descriptions, headlines, and even clean up your product photos. You can launch email and social media campaigns directly inside Shopify. And when you start scaling, Shopify scales with you. Millions of people already made the leap from first timers to brands you already know. The platform has powered it all. So if 2026 is your year to go from employee to entrepreneur, idea to income, Shopify is your launchpad. In 2026, stop waiting and start selling with Shopify.

11:12Sign up for your$1 per month trial and start selling today at shopify.com slash marketing school. Go to shopify.com slash marketing school. That's shopify.com slash marketing school. Here you're first this year with Shopify by your side. Let me show you this. Zoom info. Maybe there's two Zoom info topics this week because I think they're suing Apollo, which is one of their competitors. We can talk about that, but I also want to bring up this piece over here. Apollo is better to Zoom? Zoom info. Oh, Zoom info. I thought you meant Zoom. Yeah, yeah, yeah. Sharing. Yeah, the video. Yeah, yeah, yeah.

11:50Okay. So Henry here, who's the founder of ZoomInfo. So last quarter, we spent about$1.4 million on AI tokens, an all-time high, and the ROI wasn't what we expected. So basically, he spends$1.4 million on tokens, and he's revealing what the sales ROI is. So here's where the spend went, and here's where it actually moved to Nino. So he actually, I don't know, this is like a fake invoice due made by ChatTBD, because why would you have it all in one invoice? But anyway, I digress. So he says, number one use case for the 1.4 million in spend on AI, it's telling the reps who to call today and why. So we're using AI to sift through millions of signals and tell reps who to talk to today and why.

12:29The signals that we found matter. Job changes, so new decision makers equals new opportunities. Buying committee changes at intent signals, active web research and pricing page visits. The big ROI drivers are helping our customers with daily prioritizations so they don't have to go fishing for actionable info. So at ZoomInfo, we've seen a 25 % to 30 % increase in meeting quality and op creation when AEs are sourcing using our AI tools. Win rates also jump from 16 % to 20 % to 30%. What do you think about that? That's not bad if it's holding true, but he's saying his tool. So he's trying to pitch his tools to the market.

13:01I'm curious what other people are really seeing versus a company talking about the results from their product. because I think the results are going to be more better when he types it out versus what people see. Fair point. Number two, writing outreach that doesn't sound automated. Fair point. These things would all be built into Zoom info. So we're moving from 20 segments of 1 ,000 to 20 ,000 segments of one. Not quote unquote VP of IT at Enterprise Insurance Messaging, but John at State Farm, who we talked to last year, who competes with three of our customers with context pulled in automatically.

13:32Customer ROI here ultimately comes from better response rates and higher close rates by being more relevant. Buyers care when you show you care. I would say the things we're talking about here, Neil, with what he's talking about, is how do you apply this for your own company and maybe not look at his numbers as much because to Neil's point, his numbers are for his own product. So let me say this, okay? This is just my gut inclination. I have no data on this. And when you look at Zoom info, if their air product was helping people close 20, 30 % more, do you think their growth in their SaaS product would be crazy because people are just generating a lot more revenue?

14:09Absolutely. Okay, when I look at their quarterly revenue from September 2025 over September 2024, they had 4.74 % growth. June 2025 over the previous year was 5.25. March was negative growth, negative 1.24. December 2024 was negative 2.31. They're close to, if you average it out, close to just being flat. You know what's interesting? So I think this is the very beginning for them. I think maybe let's look at this. I think last week's video, I think we'll see a lot of this AI ROI start to come into place. Maybe 2027 will be a lot more pronounced. So I think the jury's still out because it sounds like they're just getting into this because I'm looking at number three over here and this is them using it internally.

14:55So turning sales calls into usable data, and you and I already do this, but every sales call, so ours and customers is recording using Chorus and becomes structured data. Objection patterns, competitor mentions, deal risk, coaching moments. We found the benefits of this are huge. 25 to 30 % faster ramp time for new reps and 10 to 15 % larger deal sizes through better discovery and value articulation. The average rep sells more like the best rep. And again, guys, I think in my mind, I just look at this top four as how can you apply this for yourself? I think that's the most practical thing. Go ahead.

15:24Well, I'm just looking at their stock. And when I look at their 23and3 revenue numbers is 1.24 billion. 2024 was 1.21. So a little bit of a decline. 2025 is closer to the 2023 numbers. Now, the year's not done, but it looks like they'll be closer to 2023 numbers by, call it a few percentage points. That's my rough guess. But what if they just started on this journey? It could be, but I just don't see it. they would be pushing it really hard to the public markets and creating this narrative and they would update their revenue forecasts. Because when you're a publicly traded company, you let people, you give them hints or you let the analysts know what range you think your revenue is going to be in the next quarter, the quarter after.

16:12And that starts getting incorporated into the stock price. Yep. Okay, last point over here. So I think jury's out. I think we'll see how things turn out in 2027. We'll come back to this, hopefully. Last part here is speeding up low-value engineering work. So every engineer at ZoomInfo has IntelliJ and VS Code with Klein. AI handles the unglamorous stuff, so boilerplate code, refactors, and test coverage. We've seen 25 % to 30 % faster execution on these routine tasks, which frees senior engineers to focus on system design and real product innovation. Our biggest lesson so far has been that If your data foundation is garbage, AI just helps you move faster in the wrong direction.

16:53You won't get AI working, quote unquote, until you have contextual customer prospect data centralized and you can actually build on top of it. So yeah, here we go. We're still early and we're trying to build a lot of things, but these have been the highest ROI drivers by a mile. For a lot of you, you've listened to this stuff on this podcast already, but I think it's a good reminder. These things are worth testing. That's my key takeaway from this segment. Yeah, and maybe Zoom Info is only doing it with one division of their business because they have many divisions. And if it works so well, to roll it out and then they'll start seeing more revenue gains.

17:21But I'm with Eric. And when I say I'm with Eric, I'm with Eric on the notion that most businesses are going to know better by sometime in 2027, what AI is really doing for them from a cost savings and revenue generation aspect. I think we just need a little bit more time. And on a random side note, you know, Christian's the one who got Zoom info, their money. Who's Christian? Christian Skiller from Cascadia, the investment is that the one that we met at dinner uh i don't know i don't know who you're talking about no that's another guy from cia he's a venture capitalist uh i introduced you to christian a long time he's an investment banker but he's the one who got them their money oh cool nice and it wasn't zoom info it was like discover.org or something like that they bought yeah and then they changed the name because zoom info had a stronger name in the market yeah yeah makes sense um okay so did you did you see the google rep that said ai seo is basically the same thing as traditional SEO?

18:18No, but I kind of agree. There are some nuances. But even if there's nuances, it doesn't matter. It's like saying SEO isn't the same five years ago as it is today for Google because, of course, algorithms change and you have to adapt. It's the same thing. SEO is very similar. It's just you have to adapt and pivot. Yep. So there's not much for us to add here, but I think I got this from... So Rusty Brick, you know Rusty Brick? What's it say? Barry Schwartz. Search Engine Roundtable. Search Engine Roundtable, exactly. So the tweet is now deleted, but he actually showed a picture of a guy talking at, I think Nick is one of the Google reps.

18:55And he basically said, look, it is SEO. On optimizing for AI search, Nick Fox said, it is the same as what you do for optimizing for web search and Google search. It is SEO. Build great sites and great content, he said. Nick Fox said, the way to optimize to do well in Google's AI experience is very similar. I would say the same as how you do to perform well in traditional search. And it really does come down to build a great site, build great content. And so to your point, I would say there's nuances. Every two, three years or so, SEO becomes different. And there's like a big thing that happens.

19:25But that's no different than the paid channels. That's no different than a lot of these channels. Yes, because you could say, oh, paid advertising is not the same. It isn't the same. Now you spend money on Instagram to sell products where they purchase on Instagram without ever leaving, right? You weren't doing that years ago. Does that mean Instagram is no longer, it's no longer ads? It still is. It's like SEO is. It's just adapted in the tactics and strategies you have to do. Adapt because there's multiple platforms and algorithms change. Yep. And live shopping. Okay. Oh, short form came out.

19:57Okay. Now Instagram's different, right? So these things always evolve. I think it's the people that complain that those are usually the ones that don't want to evolve. Okay, guys, that's it for today. Neil came in sick. Thank you, Neil, for doing this sick. And we'll see you guys next week. Hopefully we'll be in person. So goodbye.