Silicon Valley has been pouring money into legal AI startups. But most of these investors are not actually betting on legal.

Hey there, I’m Zach Abramowitz and I am Legally Disrupted. I’m finally back home from an 11-day road trip with stops in Colorado for ILTA EVOLVE, followed by Arizona, Napa, Reno, San Francisco, Los Angeles, and everywhere in between. It was great to be on the road and get the opportunity to meet so many of you, and I should have an update soon on my next “in-personing.”

No alternative text description for this image

Now, back to investing in Legal AI startups.

After hitting pause on legal AI investments right after ChatGPT launched, I’ve been jumping back in. Earlier this year, I invested in Sandstone (which is already showing hockey stick growth), I’m actively evaluating several other deals, and frankly, I have some serious FOMO. When I spoke to Jonathan Levy from Y Combinator last year, he told me YC doesn’t invest in verticals, they invest in the best talent. And right now, the absolute top startup talent is flocking to legal.

But it is a dangerous game out there. Yes, the products are incredible, but they’ve also never been easier to replicate, as I showed on stage at ILTA EVOLVE this week:

Just look at Cursor. Even before its acquisition by SpaceX, people were already declaring it dead despite billions in revenue because of the rapid improvement of base models like Claude Code. So what does that mean for application-layer legal companies like Harvey and Legora, especially when we’re seeing litigators at firms like Quinn Emanuel bypassing them entirely to build proprietary tools directly on top of Claude?

To figure out where the smart money is actually going, I sat down with Keith Rabois on 🎙️ episode 48 of Zach Abramowitz is Legally Disrupted. Keith who is a Managing Director at Khosla Ventures, is one of the most successful investors of the last 25 years, and, fun fact, a former litigator at Sullivan & Cromwell who clerked on the Fifth Circuit.

Keith does not pull his punches. We dug into why he is emphatically not an investor in Harvey or Legora. His contrarian take? Trying to improve the productivity of the AmLaw 100 runs completely counter to their bill-by-the-hour business model, which will just lead to fee compression rather than a massive new market. You’ll have to listen to the podcast to see which previously hot startup reminds him of Harvey.

To be fair, Keith does have a horse in this race as an investor in Spellbook. But he distinguishes Spellbook from Harvey because he doesn’t see Spellbook as #legaltech; he sees Spellbook as contract tech.

And because it’s Keith, we didn’t just stick to legal tech. We got into his prediction that the Iran conflict would be over in just three days, the massive economic boom happening in Latin America, and why he strongly disagrees with Nvidia CEO Jensen Huang about chip export controls. In fact, Keith points to the recent crash of DeepSeek as a prime example that American export controls are working exactly as intended.

It’s a fascinating, unfiltered conversation that you do not want to miss. Podcast 49 with Soxton founder and CEO Logan Brown drops this week as will podcast #50 with a very special mystery guest. So make sure to subscribe on YouTube, Apple, Spotify, or wherever you get your podcasts. Another very good option is subscribing to this newsletter.

Subscribe now

Don’t keep all the insights to yourself! Share this with a friend, peer or colleague who needs some legal disruption in their life.

Share

Tweet of the week:

Is Silicon Valley Actually Interested in…