Firm Profile
| Firm | Caruso, Yee & Partners LLP (hypothetical) |
| Size | 15 attorneys, 8 paralegals, 3 support staff |
| Practice areas | Commercial litigation, employment defense, insurance coverage |
| Location | Midtown Manhattan |
| Annual revenue | ~$9.5M |
| AI status (before) | Informal, unmanaged adoption |
The Problem
By early 2026, at least 10 of the firm's 15 attorneys were using AI tools — ChatGPT, CoCounsel, and Copilot — but nobody talked about it openly. There was no policy, no training, and no supervision framework.
What the managing partner didn't know:
- A junior associate had been drafting discovery responses with ChatGPT and submitting them without verification
- Two partners were uploading confidential settlement terms into free-tier AI tools with no data retention controls
- A paralegal was using AI to summarize deposition transcripts but had no process for checking accuracy
- Three attorneys were billing "research time" for work that AI completed in minutes — creating potential fee disputes
The wake-up call: An SDNY judge issued a standing order requiring AI disclosure certifications on all filings. The managing partner realized the firm had no way to comply — and no idea what AI use was already happening.
What They Tried First
The firm considered three options:
| Option | Problem |
|---|---|
| DIY policy — managing partner drafts something based on Google searches | No expertise in NY-specific ethics rules. Generic templates miss NYC Bar Opinion 2024-5 requirements and court-specific standing orders |
| Big Law consultants — engage a large firm's innovation team | $50,000+ for a one-time engagement. No ongoing updates. Designed for 200+ attorney firms |
| Ignore it — wait and see | 2–3 AI hallucination sanctions happen daily across US courts. NY RAISE Act takes effect Jan 2027. The risk was accelerating |
None of these solved the real problem: attorneys needed to know how to use AI safely, not just be told not to use it wrong.
The Fractal Legal Engagement
The firm engaged Fractal Legal on a Growth tier ($1,500/month) — the middle option that includes training, policy, and ongoing support.
Month 1: Assessment and Policy
Week 1 — AI Audit
- Surveyed all 15 attorneys and 8 paralegals on current AI tool usage (anonymous)
- Results: 67% using AI regularly, 0% had reviewed any ethics guidance, 40% were uploading client data to free-tier tools
Week 2–3 — Custom AI Use Policy
Drafted a 12-page firm-specific AI use policy covering:
- Approved tools list (CoCounsel, Copilot for Microsoft 365) with configuration requirements
- Prohibited uses (uploading client-identifiable data to consumer-grade AI, using AI for court filings without partner review)
- Verification protocol: every AI-assisted output requires human review and a verification checklist
- Client disclosure language for engagement letters and billing statements
- Supervision responsibilities under NY Rules 5.1 and 5.3
- Court-specific requirements (SDNY, EDNY, NY Supreme standing orders)
Week 4 — Policy Adoption
- Presented to partnership. Incorporated feedback on three provisions. Final policy adopted unanimously.
Month 2: Hands-On Training
Two training sessions (2.5 hours each, firm-wide attendance):
Session 1: "What NY Ethics Rules Require"
- Walked through NYC Bar Formal Opinion 2024-5 (the 6 core duties)
- Reviewed 5 real sanction cases with specific dollar amounts
- Live exercise: spot the ethics violation in 4 AI-assisted work product samples
- Every attorney left with a laminated quick-reference card
Session 2: "Using AI Without Getting Sanctioned"
- Hands-on with CoCounsel and Copilot against real (anonymized) case files
- Practiced the verification workflow: AI output → citation check → substantive review → sign-off
- Common failure modes: hallucinated citations, confidentiality leaks, billing ambiguity
- Paralegals included — they're often the heaviest AI users and the least trained
Months 3–12: Ongoing Support
- Quarterly policy updates — 3 revisions in the first year as new court orders and ethics opinions dropped
- Monthly court order digest — 14 new AI-related standing orders tracked across SDNY, EDNY, and NY state courts
- On-call support — 6 policy questions answered within 48 hours (most within 4)
- Mid-year refresher — 90-minute session covering new hallucination cases and updated verification protocols
Results: 12 Months Later
Risk Reduction
| Metric | Before | After |
|---|---|---|
| Attorneys with AI ethics training | 0 / 15 | 15 / 15 |
| Formal AI use policy | None | 12-page firm-specific policy, updated quarterly |
| Court-compliant AI disclosure process | None | Standardized for SDNY, EDNY, and NY Supreme |
| Client data uploaded to free-tier AI | ~40% of attorneys | 0% (policy + tool lockdown) |
| Malpractice insurance AI questionnaire | Could not answer | Completed with documentation |
Zero sanctions, zero grievances, zero client complaints related to AI use in the 12 months since engagement.
Efficiency Gains
| Metric | Before | After | Impact |
|---|---|---|---|
| Hours saved per attorney per week | 0 (informal, inconsistent use) | 4.5 hours | 3,510 billable hours recovered firm-wide annually |
| Contract review turnaround | 6–8 hours | 2–3 hours | 60% faster |
| Legal research (case law, statutes) | 3–4 hours per matter | 45 min–1.5 hours | 65% faster |
| Discovery document review | 8–12 hours per batch | 3–5 hours | 55% faster |
| Deposition summary preparation | 4–6 hours | 1.5–2 hours | 65% faster |
Revenue Impact
- 3,510 recovered hours × $350 blended rate = $1.23M in additional billable capacity
- Not all recovered time converts to billing — firm estimated $400K–$600K in actual incremental revenue from faster turnaround and capacity for new matters
- Two clients specifically cited the firm's AI governance as a reason for expanding the relationship
- One RFP win explicitly referenced the firm's AI use policy as a differentiator
Cost of Engagement
| Annual | |
|---|---|
| Fractal Legal Growth tier | $18,000 |
| Internal time (training, policy review) | ~$8,000 (estimated) |
| Total investment | ~$26,000 |
| Conservative revenue gain | $400,000+ |
| ROI | 15x+ |
What the Attorneys Say
"Before Fractal, I was using ChatGPT with one eye closed — I knew I should be more careful but didn't know what 'careful' actually meant. Now I have a checklist and it takes 30 seconds."
— Senior Associate, Commercial Litigation
"The training wasn't theoretical. They used our actual case types and showed us exactly where AI breaks. That's what made it stick."
— Partner, Employment Defense
"Three of our competitors still don't have AI policies. When we mention ours in client pitches, it's a clear differentiator. One client told us they picked us over a bigger firm partly because of our governance framework."
— Managing Partner
"I was the biggest skeptic. I thought this was just another compliance checkbox. But the efficiency gains are real — I'm handling 20% more matters without working longer hours."
— Of Counsel, Insurance Coverage
Key Takeaways
- The risk was already there. Attorneys were using AI without guidance. The question was whether the firm would manage it or get caught by it.
- Training beats prohibition. Banning AI doesn't work — attorneys use it anyway. Teaching them to use it safely produced both compliance and efficiency gains.
- NY-specific matters. Generic AI policies miss court-specific standing orders, NYC Bar Opinion 2024-5 requirements, and the upcoming RAISE Act. A template from a national vendor would have left gaps.
- The ROI case is straightforward. $26K investment against $400K+ in recovered revenue. Even at half that estimate, the return is obvious.
- It compounds. The firms that train now will be further ahead next year. The gap between AI-competent and AI-unmanaged firms is widening every quarter.
Want results like these?
Start with a free AI Readiness Assessment and find out exactly where your firm's exposure is — and what it would take to close the gap.
Start Your Free AI Readiness Assessment →About Fractal Legal
Fractal Legal provides AI training, custom policy drafting, and ongoing compliance support for NY law firms. Founded by Ansgar Lange (AI strategy and operations) and Tyler (NY attorney, legal ethics and compliance), we focus exclusively on the intersection of AI adoption and NY legal ethics requirements.
Contact: info@fractal-legal.com | fractal-legal.com
This case study is based on a composite of real market conditions, published research, and documented AI governance outcomes. The firm name and specific details are hypothetical.