DLA Piper

DLA Piper needed to move generative AI from limited late-2022 pilots into a secure, firmwide tool usable by thousands of attorneys without sacrificing accuracy or client confidentiality. Routine work such as legal research, document review, and drafting consumed significant attorney time, especially under tight M&A and matter deadlines. Scaling AI across 3,000 US lawyers also required enterprise-grade security, data sovereignty, and governance that could satisfy professional and compliance standards.

The firm completed a US-wide deployment of Harvey, an OpenAI-based platform fine-tuned for legal work, after phased pilots in contract review, due diligence, and legal research. IT partnered with Harvey to customize the system on firm documents and templates, deploy a hybrid-cloud architecture with encryption and least-privilege access, and keep US data on domestic servers. A governance framework mandates human review of AI output, confidentiality controls, ethical safeguards, and bias audits, supported by multi-tier training and practice-group AI champions.

Early data shows legal research tasks taking 40-60% less time, contract analysis completing 3-4 times faster with AI assistance, and initial document drafts requiring 50-70% less human time, while accuracy is reported as equivalent or better than traditional methods. Attorneys can ask natural-language research questions and receive synthesized answers with citations, and use Harvey to flag unusual contract provisions and generate first drafts for human refinement. Client feedback has been generally positive on speed and thoroughness, with some clients requesting Harvey’s involvement on matters.

• • Pilot carefully, but design security, data residency, and governance for enterprise scale from the start.
• • Pair tooling with training on when to use AI, prompt craft, and mandatory human verification of outputs.
• • Measure both efficiency (research, review, drafting speed) and quality/client-service outcomes when assessing ROI.