What it actually looks like when a law firm switches to AI
By Nelson Chu · Published May 12, 2026
If you handle disability, personal injury, or veterans cases, you already know the bottleneck: medical records. A typical SSD case might have 2,000-3,000 pages. A complex PI case or a veteran with 24 years of service can run 5,000-6,000 pages. Somebody has to read them, and that somebody is expensive.
AI tools like Superinsight are changing how this work gets done — handling the first pass so attorneys can focus on the legal analysis instead of reading linearly through thousands of pages. Below is how attorneys at different firm types are actually approaching this shift. These are real workflows from people we've talked to on our podcast.
Scott Haider at the Schneider Law Firm in Fargo has been reviewing medical records for SSD cases for over 13 years — he estimates his firm processes 10,000+ pages a month. For him, the medical summary is the "biggest time suck" when working up a case. His approach: use AI for the initial structured pass, then work from that output to prepare for ALJ hearings. Instead of reading 3,000 pages linearly, he reads the summary, identifies the critical sections, and goes directly to those pages.
Katie Reed at McMahan Law in Chattanooga handles both PI and disability. Her firm had a specific problem: they couldn't schedule more SSD hearings because the medical record prep was capping their calendar. Even after hiring another attorney, the bottleneck remained the same — too many records, not enough prep time.
Her take after adding AI to the workflow: "Implementing the Superinsight AI software to help us go through those medical records… has made me smarter about what I'm looking at, right? And faster." She mentioned the system flagged medical acronyms and test types she might have overlooked otherwise — not replacing her judgement, but giving her a better starting point.
For PI cases, the typical car wreck case runs about 18 months. Workers' comp can drag four years. In both cases the records pile up, and the value of a first-pass structured view is the same: you understand what you're working with before you commit hours to deep review.
Brian Mittman at The Disability Guys — a firm that's been around since 1933 — uses AI on 5,000+ page files for chronology and gap analysis before hearings. He has an engineering background, which probably explains why he described the process as: "my wheels are already off the ground" after getting the structured output. The highest use of his time as a trained attorney, he argues, is not manually sifting through thousands of pages.
His firm has handled workers' comp since the era of the Triangle Shirtwaist Fire (the 1911 fire led to New York workers' comp law, and the firm was founded in 1933 doing exactly that). SSD practice started in the late 90s. After all that history, AI record review is the first thing that meaningfully changed how prep works.
The process is straightforward. There's no onboarding call, no setup fee, no minimum commitment.
PDFs, scanned documents, faxes, images — whatever you have. The system handles OCR and document separation. No page limit, no file size restrictions. If you've got a 6,000-page veteran file, that's fine.
It identifies document types (progress notes, labs, imaging, surgical records), extracts dates, providers, diagnoses, medications, and treatments. Everything gets organized chronologically with page references back to the source.
In about an hour. Not a generic summary — a structured report organized by relevance to your case type. Every finding cites the exact page so you can verify it or pull it up during a hearing.
Tell it what to focus on in plain English: "Pull all orthopedic findings after March 2024" or "Show me medication changes related to pain managment." It regenerates the relevant sections. Unlimited revisions, same $28.
Being honest about limitations builds more trust than overpromising.
The AI structures and surfaces the medical evidence. You still decide what it means for the case, which arguments to make, and whether the evidence is sufficient. Think of it as prep work, not legal analysis.
Some attorneys (like Diane Haar) feed the output into other tools for brief-drafting. But Superinsight itself gives you the structured medical record data — what you do with it downstream is up to your workflow.
No tool does. D.K. Shillingford's approach — trust but verify on closed files first — is smart. The system catches details manual review misses, and manual review catches things the system might deprioritize. They're complementary.
As Brian Mittman's malpractice carrier would say: you're still responsible. The AI gives you a significant head start. The last 5-10% — verifying, interpreting, applying legal reasoning — is still yours.
Traditional services use nurse reviewers or paralegals who manually read the records. That typically costs $300-500+ per case and takes several days. Multiple people read your client's sensitive health information. With Superinsight, it's $28, about an hour, and no human ever sees the records. The trade-off is that you're getting AI-generated structure rather than a human reviewer's clinical interpretation — which is why attorneys like D.K. Shillingford recommend verifying against known files first.
Yes. Paul Bunn's team specifically cited the ability to read handwritten military cursive in service treatment records as a deciding factor. The system includes OCR that handles typed documents, handwritten notes, and low-quality scans. For severely damaged pages, it flags what it couldn't fully process so you know where to look manually.
The attorneys we've featured use it across Social Security Disability, personal injury, workers' compensation, VA disability, and medical malpractice. The AI adapts its analysis based on case type — for SSD it looks for RFC-relevant evidence and Blue Book criteria; for PI it traces causation from the incident date; for VA claims it identifies service-connection evidence.
Scott Haider mentioned that a malpractice-carrier CLE on AI literacy was what unlocked his firm's adoption. The key point they made: AI is a tool, and you're still responsible for verifying the output. Superinsight gives you a structured starting point with page citations so you can verify any finding. It accelerates your review — it doesn't replace the professional responsibility to review the work product before relying on it.
Correct. First case is free. After that it's $28 per case — no subscription required, no monthly minimums, no contracts. Some firms process hundreds of cases per month; some use it on one complex file when they need help. Both are fine.