Hall & Wilcox's Three-Phase AI Adoption Framework Law Firms Can Actually Use
Harvey AI published a case study featuring Steve Johns, Partner and Co-head of Technology & Digital Economy at Hall & Wilcox, on how to make AI stick inside a law firm. The framework is simple on paper and demanding in practice: start with daily usage, move to operational integration, then reimagine how services are delivered.
If you lead a practice group, run operations, or own client outcomes, this gives you a blueprint with metrics you can take to your partners tomorrow.
Phase 1: Short-term - Make AI part of the daily workflow
Goal: consistent, firm-wide usage. As Johns puts it, AI should be part of normal work, not a side project. Track active users, frequency of use, and completion of training. If lawyers aren't opening it, nothing else matters.
- Pick 3-5 priority use cases (e.g., matter scoping, first-draft emails, clause extraction, timeline building).
- Make it default: add AI checkpoints to matter intake and close-out checklists.
- Train by role (partner, senior associate, junior, PSL, BD) with short, scenario-based sessions.
- Publish a lightweight prompt library inside your DMS/knowledge hub; update weekly.
- Install "practice champions" who coach, review outputs, and collect wins/risks.
- Instrument usage: track DAU/WAU, prompts per user, and training completion.
- Set policy guardrails (confidentiality, privilege, human-in-the-loop review) from day one.
Phase 2: Medium-term - Embed AI into core operations
Goal: shift from usage stats to outcomes. Hall & Wilcox focuses on better work product, faster turnaround, and higher consistency. Crucially, use cases should support legal analysis and judgment, not just rote tasks.
- Wire AI into templates and workflows: precedents, playbooks, checklists, and model clauses.
- Integrate with your DMS, KM, and timekeeping so outputs are stored, searchable, and attributable.
- Define human review points for legal reasoning, risk calls, and novelty issues.
- Measure results:
- Turnaround time by work type (before vs. after).
- Quality/consistency scores from peer review.
- Rework rates and defect categories.
- Client satisfaction on speed and clarity.
- Hours saved vs. matter budgets (and where the time went).
- Create escalation paths for hallucinations, bias flags, or ambiguous instructions.
- Tag AI-assisted work with matter codes to analyze profitability and pricing impacts.
Phase 3: Long-term - Reimagine legal services
Goal: deliver services in new ways, not just faster. Johns points to building offerings that change how advice is produced, packaged, and accessed.
- Client-facing portals with guided workflows, playbooks, and supervised Q&A.
- Productized services (subscription regulatory updates, contract self-serve with review gates).
- Managed services for high-volume review with AI triage + lawyer oversight.
- Pricing updates: fixed fees for AI-accelerated tasks, outcome-based fees for defined results.
- Service KPIs: margin per matter, cycle time, client adoption, and expansion.
Governance that earns client trust
You're still bound by competence, confidentiality, and supervision duties. Bake these controls into every phase.
- Competence: ensure lawyers know AI's strengths/limits and supervise its use. See ABA Model Rule 1.1 comment on technology competence here.
- Risk management: establish model selection, red-teaming, benchmarking, and audit logs. The NIST AI Risk Management Framework is a solid reference from NIST.
- Confidentiality and privilege: control data flows, retention, and access; document human review.
- Procurement: vendor diligence on security, data use, and indemnities. Keep a register of approved tools.
- Client disclosures: explain how AI is used, supervision standards, and any fee implications.
How to measure ROI without a 40-page model
- Baseline now: cycle times, rework, write-offs, and client survey scores.
- Track: hours saved x blended rate, minus AI/tooling/training costs.
- Watch for second-order gains: more consistent drafts, faster onboarding of juniors, higher win rate on RFPs that ask about AI capability.
Why this matters for mid-sized firms
Hall & Wilcox shows that you don't need a thousand lawyers to do this well. What you need is clarity on phases, the right metrics at each step, and the courage to productize where your expertise is repeatable.
Clients are already using AI. They expect their firms to be at least as capable-ideally better, with governance that de-risks decisions.
30-60-90 day quick start
- Day 0-30: pick 3 use cases, publish a one-page AI policy, run role-based training, appoint champions, start tracking usage.
- Day 31-60: embed prompts into top 10 templates, integrate DMS save paths, define review gates, publish a monthly outcomes report.
- Day 61-90: pilot one client-facing service, update pricing for AI-accelerated tasks, run a governance audit, and approve the next 3 use cases.
Related resources
Bottom line: usage first, integration second, new services third. Keep the metrics honest, the governance visible, and the client impact front and center.
Your membership also unlocks:
