Freshfields is taking Claude from experiment to firm-wide infrastructure. The global law firm has signed a multi-year agreement with Anthropic to give thousands of employees access to Claude and jointly develop legal AI workflows for client work.
That is a meaningful shift for legal AI. Large firms have spent the past two years testing tools in controlled pilots. Freshfields is now treating AI as a platform layer for lawyers and business services staff across the firm.
What happened
Freshfields and Anthropic announced a multi-year collaboration on April 23. Under the deal, Freshfields is making Claude available to 5,700 employees through its proprietary AI platform, which the firm describes as a secure entry point for day-to-day legal and business work.
The agreement goes beyond normal enterprise access. Freshfields and Anthropic are setting up a co-development programme for legal-focused AI applications and agentic workflows. The target areas are familiar but high-value: contract review, document drafting, due diligence, research, and multi-step matter support.
Freshfields will also receive early access to future Anthropic models and tools. That matters because law firms are not only buying AI products; they are trying to shape how these systems behave inside regulated, high-liability work.
Why it matters
Legal work is a useful stress test for enterprise AI. Documents are long, context is messy, mistakes are expensive, and confidentiality is non-negotiable. A generic chatbot is not enough. Firms need governance, access controls, auditability, and workflows that keep lawyers responsible for the final advice.
Freshfields' rollout suggests the largest firms are moving from cautious experimentation to managed adoption. The firm says Claude is already being used daily on client matters through its internal AI platform. If that usage holds, AI becomes less of a side tool and more of an operating layer for legal service delivery.
The co-development angle is also important. Anthropic gets a demanding enterprise customer with deep legal process knowledge. Freshfields gets influence over tools that could eventually define how complex legal work is broken into AI-assisted steps.
What this means for legal tech
This kind of partnership pressures traditional legal software vendors. Firms still need trusted legal databases, workflow systems, document management, and matter platforms. But frontier AI labs are moving closer to the work itself: drafting, reviewing, comparing, summarising, and coordinating tasks.
That does not mean law firms will rip out their existing stack. More likely, the stack changes shape. Models like Claude sit inside firm-approved platforms, legal data providers embed frontier models, and firms build proprietary workflows on top of both.
The risk is overconfidence. Legal AI can make drafting and review faster, but it can also create subtle errors, citation problems, privilege questions, and unclear accountability. The firms that benefit most will be the ones that combine AI adoption with strict review standards and clear responsibility for outputs.
Our take
Freshfields' move is a signal that legal AI is becoming infrastructure for elite professional services. The interesting part is not that lawyers can ask Claude questions. It is that a major firm wants Anthropic involved in designing agentic workflows for real client matters.
That is where the value may be: not one-off prompts, but repeatable systems that help lawyers move through contract review, diligence, research, and drafting with less manual switching between tools.
The caution is just as clear. Legal work cannot be delegated blindly to a model, even a strong one. Freshfields' rollout will be judged by whether it improves speed and consistency without weakening confidentiality, judgment, or accountability. If it works, other global firms will not want to look like late adopters.
