How a Leading Law Firm Is Rewiring Legal Work With Generative AI

Australian law firm Gilbert + Tobin is moving beyond pilot projects and embedding generative AI into everyday legal production. Chief Executive Sam Nickless describes a shift where large language models handle first-pass contract review, clause extraction, and precedent search — tasks that once consumed hundreds of associate hours.

From Experimentation to Production

The firm built a secure, private environment that connects its document management system to fine-tuned models. Lawyers now prompt the system for risk summaries, drafting suggestions, and jurisdiction-specific checks. Early metrics show a 40–60 percent reduction in review cycle time for standard commercial agreements.

Nickless emphasizes that the goal is not replacement but augmentation. Senior lawyers validate outputs, while juniors focus on higher-value advisory work. The firm also offers clients a transparent "AI-assisted" billing model that reflects the efficiency gains.

Why it matters for GPU / AI infrastructure: Legal workloads demand low-latency inference, data residency guarantees, and audit-grade logging. Running fine-tuned models on dedicated GPU clusters — rather than shared public endpoints — gives firms control over confidentiality, compliance, and cost predictability.