Crimson Built AI for Litigators, Not Lawyers Who Draft Contracts

The Macro: Legal AI Keeps Building for the Wrong Lawyers

If you’ve been paying attention to legal tech for the past two years, you’ve seen about fifty AI products launch with some variation of “we help lawyers draft documents faster.” Most of them are aimed at transactional work: contracts, NDAs, corporate filings. That makes sense from a product perspective because transactional documents are structured, repeatable, and relatively easy to benchmark. You can show a side-by-side comparison and say “look, 80% faster.”

But litigation isn’t transactional work. Litigation is messy. You’re dealing with thousands of pages of case files, depositions, correspondence, court filings, and evidence that might span years. The work isn’t “draft this contract.” The work is “read everything, find the contradictions, build the timeline, draft the motion, and don’t miss the thing that loses the case.” It’s fundamentally different, and most AI legal tools aren’t built for it.

Harvey and Clio have strong positions in broader legal AI, but their litigation-specific features tend to feel bolted on rather than purpose-built. CoCounsel (now part of Thomson Reuters) made a run at legal research but stopped short of the full litigation workflow. The litigation market is enormous. BigLaw firms spend millions on associate hours doing document review alone. And yet the tooling for that specific work has been surprisingly thin.

The reason is that litigation AI is harder to build. The documents are less structured. The reasoning is more complex. The consequences of getting something wrong are more severe. A bad contract clause is fixable. A missed piece of evidence in a case file is not.

The Micro: A Former BigLaw Litigator Who Knows What’s Broken

Crimson is a three-founder team out of London. Mark Feldner is a former BigLaw litigator, which matters. He’s not guessing about the workflow. He lived it. Amine Amor and David Stromback round out the team, and together they’re building what they call the first AI platform purpose-built for complex disputes.

The product does several things that sound straightforward on paper but are technically hard to execute well. Document analysis across full case files. Visual timelines that convert thousands of pages into chronologies highlighting key events. Fact verification that compares allegations across documents and flags discrepancies. Legal drafting that reflects firm style and case specifics. Email analysis and deadline tracking for case coordination.

That timeline feature is the one I keep coming back to. Anyone who has worked on litigation knows that building the case chronology is one of the most tedious and important parts of the job. It’s usually done manually in a spreadsheet or Word document, by a junior associate billing $400 an hour to read transcripts. If Crimson can automate that with reasonable accuracy, the time savings are immediate and measurable.

They’re claiming 75% time saved on document review and say they’ve supported 500+ matters, which is a solid number for a company this young. The product integrates with iManage and SharePoint, which is important because law firms don’t switch document management systems for anyone.

Part of YC’s Spring 2025 batch, the team is four people total. They’re SOC 2 Type II certified and GDPR compliant, and they make a point of saying they don’t train AI on user data. For a legal product, that’s table stakes, but it’s good they’re saying it upfront.

Testimonials come from lawyers at Duane Morris, Milnor Law, Square One Law, and gunnercooke. That’s a mix of BigLaw and mid-market firms, which suggests the product isn’t priced exclusively for the Am Law 100. The go-to-market appears to be demo-driven through Calendly, with no self-serve pricing on the site.

The Verdict

I think Crimson picked the right fight. Litigation is underserved by AI tools, the pain is real, and having a founder who actually practiced litigation gives them a credibility advantage that most legal AI startups don’t have. The product scope is ambitious for a four-person team, but the 500+ matters number suggests they’re past the proof-of-concept stage.

The competitive risk is real but distant. Harvey has the funding to expand into litigation-specific features. Thomson Reuters has the distribution. But neither has made litigation their primary focus, and in enterprise software, focus usually beats breadth for the first few years.

What I’d watch: can they hold quality as they scale across practice areas and jurisdictions? Litigation in the UK is different from litigation in the US, and both are different from arbitration or regulatory disputes. The “complex disputes” framing is intentionally broad, and at some point they’ll need to decide whether they’re going deep in one type of litigation or wide across all of them. Both are valid strategies, but they require different teams and different products.

Sixty days, I want to see the pipeline. Are they closing mid-market firms or just demoing to them? Ninety days, the US expansion question becomes unavoidable. The biggest litigation market in the world is American, and building from London means they’ll eventually need US-based sales, US-specific integrations, and US court system expertise. The foundation looks solid. The next phase is about whether they can scale without losing the specificity that makes the product good.