OpenAI is trying to conquer the office. Legal is next.

For years, OpenAI let startups like Harvey, Legora, and others build software for lawyers on top of its foundation models — massive artificial intelligence systems trained on broad swaths of data.

Now, the company, last valued at $852 billion, is making moves to court lawyers directly.

OpenAI signaled that shift in June when it hired Jason Boehmig, an executive who spent more than a decade figuring out the slow, messy business of selling software to lawyers.

He had left a Big Law job to start Ironclad, a company that helps corporate legal departments manage contracts.

That business looks a lot sexier than it did when Boehmig spun out.

Law firms are under pressure from clients to move faster and show they are not billing armies of associates for work that chatbots can do.

In-house legal teams are looking for ways to automate more work themselves, so they can send less to outside counsel and lower their legal bills.

Suddenly, artificial intelligence has turned legal tech from what Boehmig once called an investor “wasteland” into one of venture capital’s hottest parties.

And OpenAI wants in. “Jason’s hire reflects the value OpenAI sees in the legal space,” Dan Springer, who took over as Ironclad’s chief executive last year, said in an email.

The first hints of a playbook are now emerging.

On Wednesday, OpenAI told Bloomberg Law that it is collaborating with the white-shoe law firm Willkie Farr & Gallagher to roll out ChatGPT Enterprise to every attorney and fold its large language models deeper into the firm’s own tools.

OpenAI’s legal entry is part of a broader campaign to make ChatGPT NSFW.

The company has built a desktop “super app” for software engineers, introduced secure healthcare tools for clinicians, and hired former investment bankers to train its systems in the ways of Wall Street.

OpenAI has filed paperwork to go public, and it wants to show investors that ChatGPT is not merely the thing people use to write emails, self-diagnose a rash, or ask whether their landlord can really do that.

It wants to become the place where knowledge workers do their jobs, as well as the world’s most popular chatbot.

In legal, that means taking on Anthropic, which has spent all year trying to make Claude as familiar to lawyers as a yellow legal pad.

In January, the company released a legal work plugin that could analyze contracts and track compliance, frightening investors in legacy legal software and sending shares of Thomson Reuters, Relx, and Wolters Kluwer sharply lower.

For now, little is known about what OpenAI will actually build.

In a LinkedIn post about his move, Boehmig said he would be working on products for the legal industry.

A job posting for a founding software engineer on OpenAI’s website is similarly broad, stating that the company aims to “transform legal workflows, improve access to information, and enable legal professionals and organizations to work more effectively.” OpenAI declined to comment for this story.

The question of what it builds is an existential one for legal startups.

A technology that combs through contracts for risks poses a greater threat to a business built around contracts, whereas a bundle of legal skills strikes closer to Harvey’s turf.

OpenAI was an early Harvey investor.

In interviews, Sam Altman has often described his own company as a foundation for other startups to build on, rather than a competitor with an app for every industry. “We see a future where intelligence is a utility like electricity or water,” he said onstage at a BlackRock event in March, “and people buy it from us on a meter and use it for whatever they want to use it for.” Legal may show whether OpenAI is content to sell the electricity — or whether it wants to manufacture the appliances, too.

An open question is whether OpenAI is too late to a category its main rival has already been working on. “Absolutely not,” Zach Abramowitz, a consultant who advises law firms on how technology is disrupting their industry, said in an interview.

His reasoning is simple: Lawyers may not be buying “ChatGPT for Law” yet, but many of them are already ChatGPT power users.

A recent Law360 survey found that attorneys are using AI a lot more often than they were a year ago.

The tool with the highest self-reported usage was ChatGPT, at 54% of respondents, ahead of Microsoft Copilot, Gemini, Harvey, and Claude.

In-house adoption is harder to track, but the pull may be even stronger.

Corporate legal teams have every incentive to automate more work, and many sit inside companies where ChatGPT or Claude is already being used by engineers, customer support, and everyone in between.

Lawyers were among the millions of users who road-tested ChatGTP in those first weeks after its 2022 release.

They tried asking it for legal advice (answers ranged from “surprisingly good” to “dangerously wrong”) and put it through the bar exam (it scored 50.3% correct, above the random guessing rate).

The overall results were not “that impressive for a human lawyer,” as one attorney wrote at the time, but were “pretty amazing” for a machine.

Together, these experiments pointed to something many attorneys hadn’t considered seriously before, law firm executive Oz Benamram writes in his new book, “Law, Reinvented.” Artificial intelligence was becoming good enough to take on parts of legal work.

Tech founders had their own wake-up call.

That year, Harvey’s founders cold-emailed Jason Kwon, OpenAI’s general counsel at the time, convinced that the models could be tuned to make legal research, drafting, and other lawyerly tasks ridiculously faster.

Today, Harvey serves more than 100,000 legal professionals and is valued at $11 billion.

Harvey became the proof point for a bigger gold rush.

From San Francisco to Stockholm, founders raced to package large language models as apps for lawyers.

Funding to legal tech startups worldwide reached $2.1 billion in the first half of 2026, per Crunchbase data.

Anthropic jumped headlong into the category.

In May, the rival lab released Claude for Legal, a library of plugins aimed at law firms and in-house legal counsel.

It targeted the grunt work of the profession: finding cases, reviewing contracts, and routing documents and requests.

Meanwhile, the company that helped create the legal tech boom largely sat out the next chapter.

OpenAI has sold ChatGPT Enterprise to a growing number of law firms, and last year, it blogged about an internal tool its employees used to spot issues and pull information from contracts.

But that system was built to help OpenAI run its business and not released as a product for lawyers.

Boehmig joined OpenAI seven weeks ago.

He’s already putting a band together.

In July, Stacey Wang left Ironclad for OpenAI to lead industry marketing for legal and professional services.

In social media posts about their moves, both said they’re working closely with Sherwin Wu, who runs engineering for B2B Verticals at OpenAI Platform, the business that lets other companies build products using OpenAI’s models.

OpenAI has also been staffing people to answer a more basic question: Are its models actually good enough for legal work?

In April, the company hired a former corporate attorney as a legal evaluations specialist.

AI labs use evaluations to measure how well their systems perform on real-world tasks, so they can tell whether each new version is actually getting better.

The hires have sharpened a fear that has been building across legal tech.

The model companies may not be content to power everyone else’s software.

Winston Weinberg, Harvey’s founder and chief executive, is already used to living with the tension.

Even before OpenAI began staffing up for legal, Weinberg told Business Insider that Harvey would always be compared with whatever broad-use tool OpenAI shipped last.

On a recent call with Business Insider, Weinberg seemed coolheaded, even as Harvey prepares for a future in which one of its most important partners could become a more direct challenger.

His view is that every new entrant helps Harvey by drawing more attention to legal AI and showing lawyers what the technology can do.

His message for OpenAI: “Welcome to the party.”