Insights / The Future of Medical-Legal Services

Building an AI-Ready Personal Injury Practice: Where to Start

By Breen LNC.

Short answer: Start small and in order. Find out what your team already uses, write a one-page policy, pick one use with a clear payoff such as medical records, vet the vendor, pilot on closed files, and assign a qualified person to check the results. Then measure and expand.

Step 1: Find out what’s already happening

Your people may already be using these tools. 30% of responding lawyers said they used AI in the ABA’s 2024 survey, as reported by the ABA Journal. Ask without blame, because you can’t manage what you don’t know about.

Step 2: Write down a simple policy

A 2025 survey by Thomson Reuters, which sells AI products, found only 22% of professionals in legal, tax, accounting and other fields said their organization had a visible, defined AI strategy. Yours can start on one page:

  • Which tools are approved, and for what
  • What client information may go into each
  • Who must check output before it is relied on
  • Who to ask when unsure

Step 3: Learn your duties

Read your state’s guidance. For examples, Washington’s opinion covers competence, confidentiality, client communication, fees, candor and supervision. It says sensitivity may call for consulting the client and sometimes getting informed consent, and that an engagement agreement is a logical place to record that conversation. Arizona’s guidance quotes ABA Opinion 512 that lawyers should read and understand the terms of any generative AI tool. These are advisory, even in the states that issued them, and they show what to think about. Check your own state’s rules and opinions.

Step 4: Choose one problem

Pick something with volume and a clear measure. Medical records are a natural fit for an injury firm. Choosing a single use also means you can tell if it’s working.

Step 5: Vet the vendor

Ask about data use, retention, subcontractors and security evidence. Ask who reviews output and how accuracy was measured.

Step 6: Pilot on closed files

Run cases you know. Count errors and omissions separately. NIST’s AI RMF core calls for testing before deployment and regularly during operation, and tracking risks over time. It’s voluntary, and it’s good discipline.

Step 7: Assign responsibility

NIST’s framework says human roles in overseeing AI systems need to be clearly defined. Put a name beside each task. Legal output goes to an attorney. Medical facts go to someone with clinical training, on staff or outside the firm.

Step 8: Train, measure, adjust

Teach people how the tools fail, not just how to use them. Track a few numbers, such as hours per case, errors caught and time to demand. Revisit the policy regularly, because ABA Opinion 512, as quoted by Washington’s committee, calls AI tools “a moving target.”

What you can skip for now

You don’t need a large budget, a dedicated AI team, or a tool for every department. Starting with records and a qualified reviewer will teach you most of what you need to know. Breen can be that reviewer.

Frequently asked questions

Where should a personal injury firm start with AI?

Find out what staff already use, write a short policy, choose one use such as medical records, vet the vendor, pilot on closed files and assign a qualified person to review output.

Do lawyers have to tell clients they use AI?

It depends on your jurisdiction and the circumstances. Washington’s advisory opinion says sensitivity may call for consultation and sometimes informed consent, and Arizona’s guidance quotes ABA Opinion 512 that informed consent is required before putting client information into self-learning tools. Check the rules that apply to you.

Sources

General information for legal professionals. This is not legal or medical advice.