AI in legal departments: Lessons from ELM Amplify 2025
Legal * December 5, 2025
Over the past year, AI in legal has moved from theory to daily practice. Many teams are past informal pilots and are building real strategies. That shift was front and center at ELM Amplify 2025, where leaders from Marsh McLennan and DHL shared what actually works-and what slows you down.
Their message was clear: success isn't about chasing the newest tool. It's about focus, governance, and solving the right problems for your organization.
Why a focused strategy matters
A focused strategy turns AI from one-off experiments into a core part of the department's plan. It helps you pick high-impact problems, align work with business goals, and avoid spinning up tools that won't scale or last.
The adoption strategy that works
Pursue quality of use cases, not quantity. Smaller, well-scoped projects win more often and get adopted faster. Tiffani Huynh (DHL) noted her team trimmed a long initiative list to a tighter, long-term plan tied to legal's goals. Chris Terry (Marsh McLennan) emphasized staying locked on core departmental needs to avoid vendor noise and scattered internal requests.
The filter: will this solution be effective, durable, and worth maintaining? Build a clear plan, cut distractions, and prioritize the few initiatives with the biggest organizational benefit.
How legal teams are using AI right now
Leading departments have moved from pilots to live tools that save time and expand capability. Here's where progress is real:
- Marsh McLennan
- Solid AI governance process in place
- Practical adoption via AI features in existing external solutions
- Translations
- Pre-execution contract comparisons and redlining
- M&A due diligence and routine legal admin work using an in-house ChatGPT-based platform
- DHL
- EEOC position statement generator
- Contract extraction and drafting
- Shift from an in-house tool to Copilot, with legal-specific prompt training
- Legal operations career architect agent
- Pilots for document review, contract creation, and analysis
In an audience poll, Copilot was the most-cited AI tool-foundational platforms are gaining broad traction.
What slows AI adoption
The biggest blockers are inside the organization: approvals, governance, and risk management. Large enterprises move slowly, and data sensitivity adds friction. Multi-layer approvals and new AI committees can stall momentum, disengage stakeholders, or make solutions obsolete before go-live.
To keep things moving, Huynh recommended steady internal marketing and storytelling. Support from IT and finance is essential, but those teams are stretched. Bring them into roadmap talks early to secure alignment and resources. Change management is a constant, so plan for it explicitly.
How to overcome resistance
Pair top-down support with bottom-up momentum. Get leadership to sponsor priorities and standards. Then activate internal advocates to run demos, share wins, and answer real concerns. Clear communication plus visible value builds trust fast.
Make steady progress: three moves to prioritize
- Stop chasing shiny objects. Build a simple strategy that maps to your legal and business objectives. Pick problems with broad impact and high frequency.
- Anticipate internal barriers. Engage IT, security, and finance early. Show a tight business case and a clear governance model (consider anchoring to frameworks like the NIST AI Risk Management Framework) to speed approvals.
- Start with mindset, then metrics. Drive adoption first and collect qualitative feedback. Use quick, visible wins to justify deeper investment before you chase hard ROI.
Practical next steps for legal leaders
- Pick one use case per quarter and ship it. Contract playbooks, redlines, and intake triage are strong starters.
- Stand up lightweight governance: data access rules, model usage guidelines, review and audit checkpoints.
- Create an internal AI champions group. Share templates, prompts, and results in a monthly show-and-tell.
- Stay close to compliance signals (for example, the EEOC's AI guidance) and document your control decisions.
If your team needs structured upskilling, explore role-based learning paths here: Complete AI Training - courses by job.
Your membership also unlocks:
