Beyond the Hype: How PI Firms Can Adopt AI with Confidence

Watch the full webinar recording here →

Host: Jim Sinai, CMO of Supio

Guest: Eric Sanchez, Maestro Strategic Partners

Meet Our Panelist

Eric Sanchez

• Law firm operator for ~20–25 years, including a $1.25B civil rights class action.
  
• Invited to the White House for that work.
  
• Founded and exited an intake and case management software company.
  
• Holds 30 patents related to email search.
  
• Now helps plaintiff firms evolve analog workflows into AI-native operations.
  

Jim: You’ve watched AI for a long time going back to your 2018 ABA presentation. With ChatGPT exploding, do firms need PI-specific AI or can they use generic platforms?

Eric: It’s like practice itself. A PI-focused firm is a better option than a generalist. Same with AI. A closed-circuit, vertical AI built to advance plaintiff firms is a no-brainer. And the ethical implications of open systems are legion – Sam Altman recently said your queries might be discoverable. If firms are throwing medical records into general GPTs: buyer beware.

Jim: What matters most about building AI into workflows?

Eric: Efficiency and value. PI firms are contingency-based: how fast can we move and how much value can we maximize for clients and the firm? Many people love AI but don’t think through where to apply it.

• Litigation-first shops (90%+ litigation) should use AI to take the pain out of litigation.
  
• Pre-lit-heavy shops (90/10 pre-lit) should focus on medical chronologies, demands, and lightening paralegal burden so they can handle more cases.
  

Jim: Slick demos are everywhere. Where does AI fall flat and how do you know when it’s real?

Eric: Some features are commoditized (like basic case Q&A). The test is the end product: how does that letter look? What’s that medical chronology really doing? Where is time actually saved? Don’t buy the sexy, buy the practical. Be wary of “99% accuracy”. There’s no perfect paralegal, so keep human-in-the-loop with belts and suspenders. If a company doesn’t want questions or competitor comparisons, take a second look.

Jim: We tell buyers our AI is 95% accurate because we benchmark it – and to set expectations.

Eric: I hear the honesty of the five, not the delta to 100. That builds trust and avoids malpractice thinking.

Jim: New features roll out constantly. Jump in or wait?

‍Eric: I’m a measure-twice-cut-once guy. I probe the underlying tech, the practical value, and whether it scales. Sometimes the immediate ROI says jump in; other times beta internally first. What matters is vendors that listen. Too many think they know the problem and dictate solutions. I love Supio’s Prompt Library: take a best-practice prompt and scale it; let everyone benefit when it’s improved; make quality evenly applied across matters.

Jim: How is AI moving the needle for firms?

Eric: Two angles:

1. Settlement value. In Colorado, during a depo, the defense said the client never complained of injuries. A quick query surfaced 14–15 prior complaints, turning a six-figure case into a $3M case.
   
2. Time on desk. Benchmark how long soft-tissue and other cases take from open to resolution; how long medical records and demands take. Without benchmarking, you’re leaving money on the table.
   

Jim: Ultimately, it’s profits – can you do more cases with less resources?

Eric: Pair great paralegals and support staff with AI and you can have confidence in outcomes while helping the bottom line.

Jim: Adoption is hard. What separates firms that succeed?

Eric: Great firms rest on people, process, technology. Resistance to change is real – especially with AI. Build internal champions of innovation who ask how can it work (not why it won’t). Get them testing early and even writing an internal SOP. Incentives exist, but the spirit should be client value. If someone’s a fringe employee, AI isn’t the problem – the work ethic is. The upside is people spend more time with clients (and family) and less on grunt work.

Jim: Biggest implementation mistake?

Eric: Believing the pitch without hands-on testing. Get paralegals and staff in early; they catch details like decimal points on financials (33% vs 33.33%). The purchaser writes the check; staff live in the product.

Jim: Future: what’s next?

Eric: Agentic AI and cross-case analysis.

• Agentic: e.g., balance checks – have an agent reach out to providers, retrieve results, and embed them automatically. Paralegals hate sitting on hold.
  
• Cross-case: huge for mass torts/class actions – see, say, 67-year-old diabetic clients on a certain drug, then act on it. Even predict how Allstate will move and where settlement value may land with enough data.
  

Jim: That’s the Iron Man model – AI as the suit.

Eric: Exactly: great paralegals become superheroes; great attorneys become super attorneys.

Audience question: How do you come up with settlement values if they’re confidential?

Eric: They’re confidential externally, but internally you should capture them in the record and in your database.

Audience question (Christina): How long to see ROI?

Eric: It depends on philosophy. Some measure by reducing headcount (slower to calculate). If you want case value up 10%, you likely need to see the full case cycle (12–24 months). For time on desk, you may see it in a quarter. If you mean $1 in, $5 out, you need a full life of a case.

‍Where to reach Eric: mstratpartners.com.