Since November, he has been churning out drafts of merger and acquisition agreements as well as memos to clients. He never leaves the office and never takes a coffee break. He’s also not human.

Harvey is the name of the artificial intelligence chatbot the international legal giant has been testing for the past several months, without informing its clients, the Financial Times reported. The tool is available to any of the company’s attorneys.

It’s the latest expansion of A.I. as a workplace tool, and one that has raised concerns given the sensitive nature of legal contracts.

Allen & Overy tells the FT that any contract Harvey creates is fact-checked by company attorneys—and the purpose of the tool is to create baseline versions of documents, which the lawyers can use as a template to improve upon.

“It’s not cutting out anyone, it’s not a cost-cutting exercise, it’s a nice smart way of working,” David Wakeling, head of the firm’s markets innovation group, told the paper. “It’s saving time at all levels.”

While A.I. tools are being used by everything and everyone from search engines to teachers, Harvey is the first known example of a top London law firm utilizing it. Allen & Overy underscored that the A.I. system is not replacing any of its workforce, nor will it reduce billable hours.

It also won’t save clients any money in the near future, though the firm says cost savings could come as the technology gets more advanced.

While A.I. chatbots might be making legal headway in the U.K., the U.S. is a bit less welcoming. Last month, DoNotPay announced plans to bring “the world’s first robot lawyer” into a court case, having an A.I. chatbot “listen” to the case and generate responses, which would have been repeated back to the judge. The company backtracked on that plan after the CEO said he believed he would be jailed for six months if he did so.