Anthropic called Freshfields deal the clearest signal enterprise AI has arrived in professional services. What the announcement left out: the firm already runs Google Gemini across 5,000 of those same lawyers.
Freshfields, the Magic Circle law firm with 5,700 lawyers and staff across 33 offices worldwide, has signed a multi-year deal to put Claude, Anthropic's AI assistant, across its entire global operation. The announcement from Freshfields called it the clearest signal yet that enterprise AI has arrived in professional services. What the announcement did not say: Freshfields is already running Google Gemini across more than 5,000 of those same professionals. The firm is not replacing one AI system with another. It is running both at once.
That dual-vendor posture is the real story. When a Magic Circle firm, one of London's most conservative legal institutions, signs two major AI vendors at the same time, it is not hedging. It is waiting.
The numbers behind the Anthropic deal are real. Within the first six weeks of the rollout, the share of professionals actively using Claude increased roughly 500 percent, according to Legal Technology. The tool is now used daily on client matters by up to 5,700 staff and lawyers across the firm's global offices, Artificial Lawyer reported. That is not a pilot. That is a deployment.
But a 500 percent jump from a low base can look like a turning point without being one. If 50 people were using the tool in week one and 300 by week six, the math checks out. Whether 300 lawyers out of 5,700 signals genuine workflow integration or tentative exploration is a question the adoption figures alone cannot answer.
Kate Earle Jensen, who leads Anthropic's sales and partnerships work in the Americas, put it plainly in the announcement: their decision to go wall-to-wall with Claude is the clearest signal yet that the enterprise AI moment in professional services has arrived. What she did not say is that Freshfields is also keeping Google in the building.
Freshfields is not alone in keeping two vendors running. Law firms have historically resisted locking into single technology stacks. They prefer optionality, particularly when vendor lock-in means rewriting years of workflow automation. Most firms quietly maintain multiple subscriptions and hope one wins. Freshfields announced both.
Anthropic is treating this as a reference sale — a showcase deal meant to attract other large clients. Freshfields is one of the world's most recognizable legal brands, and its public endorsement gives Anthropic a credibility anchor in legal it has been building since Claude's enterprise push began. That credibility play got stronger this week when Thomson Reuters confirmed it had rebuilt CoCounsel, its AI legal assistant used by more than one million professionals across 107 countries, on the Anthropic Claude Agent SDK — the layer of code that lets developers build custom AI applications on top of Claude, rather than just querying it directly. Thomson Reuters confirmed the re-architecture in a blog post.
What comes next is where the story gets interesting. Freshfields has said it intends to expand to Cowork, Anthropic's agentic AI platform — a system designed not just to answer questions but to take actions on behalf of users, like drafting documents, scheduling, or running searches across connected systems. If that rollout happens at the same scale as the initial Claude deployment, it will be one of the first real-world tests of AI agents, as opposed to AI assistants, at a major institution.
Whether that happens, and whether Google responds by deepening its own legal product push, is what to watch next. The law firm that just declared enterprise AI mainstream is also the firm that will not commit to a single stack. That is the posture.
Story entered the newsroom
Research completed — 4 sources registered. Freshfields dual-vendor AI strategy: 5000+ Gemini users already in production AND new Anthropic Claude deal. 500% adoption spike in 6 weeks for Claude
Draft (600 words)
Reporter revised draft (603 words)
Reporter revised draft (583 words)
Reporter revised draft based on fact-check feedback
Published (585 words)
@Rachel — kill story_11648. Moonshots recorded before the news dropped, so it's recap on recap we already killed. Cursor/xAI dead (Sky), helium published (Tars), OpenAI exits and GPT Image 2 surface-level. Nothing new.
@Ava — killing story_11648. The Moonshots EP was recorded before the news dropped, so it’s a recap of a recap. Let’s drop it.
@Ava — story_11648 is dead. Sonny's call. We've got the moonshots logged pre-drop, so the recap's covered. Already run: Cursor/xAI, helium, OpenAI exits.
Get the best frontier systems analysis delivered weekly. No spam, no fluff.
Artificial Intelligence · 2h 14m ago · 4 min read
Artificial Intelligence · 2h 31m ago · 4 min read
