|
24 August 2026 (Washington, DC) - Anthropic’s in-house legal team scales its lean department by using its own AI tool, Claude, as an operational "force multiplier" to automate repetitive legal workflows.
They dogfood their own technology to handle explosive corporate growth, which ultimately led to the public release of Claude for Legal.
The team operates on the philosophy that AI cannot replace legal judgment, but it can scale it. Before automating tasks, Anthropic’s lawyers explicitly codified their expertise and playbooks into structured guidelines.
Anthropic's specialized legal tech expansion has fundamentally shaken up the legal software market.
And it is growing exponentially. Much has been written about the IPO battle between OpenAI (a fading star) and Anthropic, which filed privately for a proposed initial U.S. public offering this summer (possibly going public this year) and it hired a lead for its popular suite of legal tools. OpenAI was also supposed to launch an IPO this year but has now pushed back "possibly" into 2027 due to a raft of problems which we'll cover in a separate post.
But before all that Anthropic had a product lawyer moonlighting as a bit of an engineer inside the legal department.
Associate general counsel Mark Pike, who was hired by the company three years ago to lead a team of product shepherds, said in an interview last week that he had been relying on his rule of thumb as an in-house product attorney when he ended up helping the company create and launch the expansive AI tool set known as Claude for Legal. His operating principle: "I need to know the product well, and the best way to learn is to use it".
Pike, previously a product lawyer for the messaging app Slack, began experimenting with Anthropic's chatbot, Claude, to see how it worked on his own legal tasks. Last year, the company's sales team tapped him to start meeting with in-house legal departments at other companies that were looking for tips on how to use it.
Eventually the technology developed so that Pike was able to build tools "that really worked for lawyers," he said.
The company decided to launch them. Suddenly, Pike was no longer just a product lawyer, "but also a bit of a product lead or a product manager," he said.
Claude's legal plug-in arrived in February 2026, and more than 20,000 people registered for an April webinar on how to use it - the company's largest-ever session, it said. The following month, Anthropic released a dozen practice-area-specific plug-ins, all open source, and more than 20 connections to other legal technology tools.
The San Francisco-based company confidentially submitted a proposed initial public offering to the U.S. Securities and Exchange Commission in June of this year, just days after it hit a post-money valuation of $965 billion after securing $65 billion of investor commitments in its massive Series H funding round. This month, it hired Harvard Law grad Robert Mahari, who founded the AI engineering firm Akiva AI, to head up Claude for Legal.
Pike said he expects to remain in the mix, "advising and serving in some capacity," with Mahari at the helm. His experience helping bring Claude's legal tools to life is one that's increasingly common in the AI space, he said, where lawyers are learning by building.
Internally, Anthropic embraces a culture of "antfooding," Pike said, a play on the term "dogfooding" - the practice of using your own product and testing it in the real world - with employees being the "ants". He said:
"We use our own products to see if they work well, with the legal department often giving feedback to product managers and builders".
It was the legal department that suggested changes to Claude's understanding of legal formatting and how to edit legal documents, Pike said. And, after the legal tools were released, it was Pike who noticed that people were having a hard time finding them in the larger directory and suggested permalinks.
As other in-house departments, including their legal operations teams, work toward AI adoption, Pike shared what he considers three innovative ways Anthropic's legal department is using its own tools to help them with legal work:
- Pike built an internal tool using Claude that detects the product team's calendar entries for new product launches and automatically starts reviewing them and providing initial feedback based on the legal department's playbook, or rules for review.
"In the before times, product lawyers would have to scan Slack messages or follow along in a spreadsheet," Pike said. The tool also flags the attorney covering the launch, to keep a human in the loop.
- The commercial legal team built a model context protocol that allows Claude to connect to the existing contract management system, Ironclad. Instead of copying data in and out of programs, the tool can read contract information where it already sits, Pike said.
"Before, someone had to pull contract language out of Ironclad, run it through Claude separately, then paste results back in," he said. Now Claude reviews redlines and drafts edits, "so, a lot of the first pass back-and-forth happens immediately, and lawyers spend more time on meaningful work."
- The legal team uses a tool called Claude Tag to get a head start on reviewing marketing materials when the turnaround is tight for new product launches.
"Oftentimes we were getting pinged by our marketing team the day before launch that they need us to read a blog post," Pike said.
The AI tool allows the marketing team to tag the Claude agent in their messaging with the team's attorneys, giving it access to their rules for review, such as the top 10 issues they frequently spot in marketing announcements, Pike said. It's another instance where the AI can do a first pass on attorney review and cut down on turnaround time, he said.
While not a substitute for a human lawyer, who must still review and sign off, "what it does is give our lawyers a head start," Pike said. "By the time an attorney looks at it, some of the routine triage is already done".
| 
