AI Legal Tech Is Saving Hours. Firms Are Wasting It.
Document review used to be the thing that killed law firm margins. A single complex case could demand weeks of junior associate time at $200-400 per hour, manually reading through thousands of documents to find the needle in the haystack. That was the business model. That was the problem.
Then AI showed up and started solving the problem. And that's when things got weird.
Tools like Relativity aiR, CoCounsel Legal, Luminance, and Kira Systems are genuinely fast. The numbers are there: Relativity's work with Microsoft yielded projected annual cost savings of $1.6 million. CoCounsel delivers 63% faster document review and helps lawyers find twice as many relevant cases in the same timeframe. Law firms with formal AI strategies see 3.9x higher ROI than non-adopters.
But here's the problem: None of that matters if you can't bill for it.
The Math That Doesn't Work
Law firms have a billing problem. And it's not new — it's just been exposed by AI.
Ninety percent of legal dollars still flow through hourly billing. A firm deploys a $50,000 AI tool that cuts document review time in half. An associate now completes in 20 hours what used to take 40. The client saves 20 billable hours. And then what? The firm either:
Option A: Bill the client for 20 hours of work that AI mostly did, knowing the client will eventually find out and lose trust.
Option B: Bill for the actual 20 hours, pocket the margin difference, and hope clients don't demand rate cuts when they realize the work is faster.
Option C: Propose value-based pricing, watch the procurement team convert it back to hourly rates anyway, and abandon the experiment.
Most firms are picking Option B and hoping nobody notices. That's not a sustainable strategy when corporate legal departments have been using gen AI since 2022 and know exactly what the efficiency gains look like.
Thomson Reuters' 2026 State of the Legal Market report calls this "an almost absurd tension." Firms are investing at record rates — tech spending surged 9.7% in 2025, the fastest growth since the 2008 financial crisis — but they're deploying tools into a billing structure that punishes efficiency. The report notes that "clients aren't eager to see all their productivity benefits flow straight to law firm profits. Nor are they prepared for the sticker shock of a $2,000 hourly bill from an associate."
The GCs already know what's happening. They're just waiting to see if law firms will figure it out.
The Accuracy Problem Nobody's Talking About
There's another reason to be skeptical. These tools work, but they hallucinate.
Stanford researchers tested the major legal AI platforms and found that Westlaw's AI hallucinates at a 33% rate while LexisNexis's Lexis+ AI hallucinates at 17%. That means roughly 1 in 3 to 1 in 6 answers are confidently wrong. In law, confidently wrong is worse than slow.
This is why the real workflow isn't "use AI to do legal research." It's "use AI to do legal research, then have a human lawyer verify everything." The time savings shrink. The liability risk doesn't.
CoCounsel, Luminance, and Kira are better at this because they're designed for specific tasks — contract review, document analysis, due diligence — where hallucinations are easier to catch and the stakes are slightly lower. But even there, you need human review. The 63% speed improvement assumes someone still checks the work.
Who's Actually Winning
The firms getting real ROI aren't the ones betting on AI to replace lawyers. They're the ones using it to let lawyers do lawyer work.
A mid-size firm using Relativity aiR or CoCounsel isn't trying to bill fewer hours. They're trying to handle more cases with the same number of people. They're using the time savings to take on clients they couldn't afford to serve before, or to shift junior associate time from document review to actual legal analysis. The efficiency gain becomes a competitive advantage in client acquisition and retention, not a margin squeeze.
Small firms especially are seeing this work. A 10-person firm that adopts AI legal tools effectively can compete on turnaround time and cost with firms 10x their size. That's real value. But it requires intentionality — a formal AI strategy, not just a subscription.
The firms that aren't winning? The ones that bought the tool, deployed it, and expected the billable hours to stay the same. (Spoiler: clients notice.)
The Billing Model Problem Is Unsolvable Without Change
Here's what needs to happen: Law firms need to stop pretending hourly billing works with AI. It doesn't.
Value-based pricing, fixed fees, outcome-based arrangements — these aren't new ideas. But they require firms to actually negotiate with clients instead of just applying a rate card. It requires transparency about how AI is being used. It requires procurement teams to stop converting everything back to hourly equivalents.
Legal tech spending hit record growth in 2025, but the billing model hasn't moved. That's a collision waiting to happen. Either firms will figure out how to price AI-enhanced work differently, or clients will keep demanding rate cuts to capture the efficiency gains themselves.
The technology is doing what it's supposed to do. The business model is broken. Until that changes, "AI is saving us time" just means "AI is making our margins worse."