
Legal AI & Technology
Plaintiff Firms Say They've Adopted AI. Their Own Numbers Say Otherwise.
By Sagar Pratap SinghPublished Last updated
Supio's survey of 207 PI attorneys: 78% have engaged with AI, only 30% use it regularly. The blocker is trust, not price: 99% won't use output they can't verify.
Full issue
Most personal injury firms have tried AI. Fewer than a third actually run on it. Supio's 2026 survey of 207 U.S. plaintiff attorneys and firm leaders found 78% have engaged with AI in some form, but only 30% use it regularly. Another 23% are experimenting on isolated tasks, 25% are still evaluating, and 22% have no plans at all.
The gap between "tried it" and "trust it" is the real story on the plaintiff side. It is an operations problem and a verification problem before it is a technology problem, and the firms that close it will be the ones that make vendors prove accuracy on their own files.
The barrier is verification, not price
Supio's headline finding is that trust, not cost, is holding firms back. The numbers behind that are blunt: 99% of respondents said they will not use AI-generated content they cannot verify, and 96% are very or extremely concerned about output that can't be traced to a source.
Ask why five times and you land in the same place. Firms stall because they don't rely on the output. They don't rely on it because they can't check it quickly. They can't check it because many tools hand back a finished paragraph with no page cite into the medical record. So the attorney re-reads the file anyway, the time saving disappears, and the tool becomes a line item nobody defends at renewal. A discount gets a firm to sign. Only traceable, correct output on its own files keeps it using the product. That is why 71% of respondents ranked source-level traceability as a priority, level with accurate medical record summaries.
The pressure is real, and it is operational
Firms are not shopping for novelty. Three-quarters say they are at or beyond capacity. 42% say case volume regularly exceeds what the team can handle, and 57% need a week or more to get a demand letter out. On most PI matters, the demand is what starts the money moving, so every extra week sits directly between a signed client and a settlement.
That explains why plaintiff-focused vendors lead with record summarisation and demand drafting instead of general chat. It also explains who converts first: the buried firms, not the curious ones.
Everyone is early on proof, plaintiff firms just feel it faster
Corporate firms are further along on rollout. Thomson Reuters' 2026 professional services report puts organisation-wide AI use at 40%, up from 22% a year earlier. But rollout is not the same as results. Only 18% of organisations track AI ROI, and Litify's 2026 research found fewer than 15% of legal professionals see measurable business impact. Clio's 2025 Legal Trends Report adds another wrinkle: use of legal-specific AI tools fell to 40% from 58%, as people drift to general-purpose assistants.
Plaintiff firms feel the proof problem more sharply because their work product is case-specific and their margin rides on it. A demand with the wrong policy limit or the wrong date of injury does more damage than a slow one. After 14 years selling into these firms, the pattern I see is consistent: the firm that moves from the 78% into the 30% is the one that watched the tool get its own files right, with a citation it could click, before signing its name to the output.
What to do this week
- Pull three closed files with known outcomes and run them through any tool you are evaluating. Score it on errors in dates, policy limits, providers and billed amounts, not on speed.
- Make "show me the source page" a pass/fail test. If an extracted fact doesn't link back to the record, treat the output as a draft, not work product.
- Measure your current demand turnaround from records-complete to demand-sent. Without a baseline, you will never know whether AI paid off. Only 18% of organisations track AI ROI at all.
- Name one owner for AI rollout with a 60-day target, so "experimenting" has an end date.
Sources
- Supio: Trust, Not Cost, Is Blocking AI Adoption in Plaintiff Law (PR Newswire, July 2026)
- Supio: How Many Plaintiff Firms Actually Use AI? (2026 Data)
- Thomson Reuters: 2026 AI in Professional Services Report
- Litify: State of AI in Legal (Business Wire, Feb 2026)
- 2Civility: 2025 Clio Legal Trends Report summary