In-house legal teams get creative with AI tools - Financial Times

In-house legal teams get creative with AI tools

Roula Khalaf, Editor of the FT, selects her favourite stories in this weekly newsletter.

Rather than buying a legal-AI tool, Kensuke Tsujimoto is creating his own.

The top lawyer at Japan-based fintech Upsider is refining a tool that feeds conversations on Slack, Notion and other platforms to Anthropic’s AI model Claude alongside a legal playbook developed in-house. Claude highlights potential regulatory and contractual red flags, as well as instances where Upsider’s legal positions may need to be updated.

Tsujimoto then reviews the possible conflicts. “These questions come to me to determine if this is a false alert or if this is actually a real risk,” he says. “Then we can step in and help our business [colleagues] even before they come to us for legal advice.”

Legal teams at companies of all sizes are trying to be creative in their use of AI, and experimenting with new tools to speed up their work, make time for more meaningful projects and minimise risk. In some cases, they are using the technology in ways the developers did not envision.

Cox Media Group, which operates TV and radio stations in the US, discovered it could adapt a contract-processing tool to streamline its reviews of political ads, a process that is “full of potential friction points,” says general counsel Eric D Greenberg. While those ads are an important source of revenue — total advertising spending in the 2026 US midterm elections is forecast by tracker AdImpact to reach a record $11.6bn — they must be evaluated carefully to avoid defamation risks.

“There are products that are quite reasonably and logically marketed for very broad addressable markets, where you’re able to create an ‘off-label’ use that may be idiosyncratic or bespoke,” he says. “It can make a huge difference.”

Nanyang Technological University in Singapore has used a contract platform from Lexagle to strengthen its response to reports that allege wrongdoing and misbehaviour across the institution, says Greg Chew, its chief legal officer.

The tool has helped accelerate investigations, shortening the average time needed to close a case to 38.5 days, down from 55 days, he says. In late 2024, more than half of such reports were flagged for investigation, but a year later more efficient assessment resulted in less than a third needing further examination.

Elsewhere, tech services company Endava has used OpenAI’s ChatGPT to build customised tools to tackle some of its “pain points”, from evaluating non-disclosure agreements to screening customers, suppliers and others, says Rohit Bhoothalingam, its general counsel.

“Some lawyers are wary of the technology,” he adds. “Others are good at it, but use it primarily for their own efficiency gains and the team doesn’t benefit — the company doesn’t benefit.”

Regular hackathons, where a representative from OpenAI has joined Endava staff, aim to overcome those challenges, spurring collaboration and allowing the company to adapt as AI evolves. In one example of the progress it has made, the company estimates that the time required to screen third parties has declined by 40 per cent to 50 per cent, despite an increase in requests.

Bhoothalingam says his team is more productive while keeping headcount stable at about 35 people.

Upsider, majority owned by Japan-based Mizuho Bank, is also keeping lean. It plans to expand from two in-house lawyers to just four while its revenue grows at a rate of about 50 per cent a year, Tsujimoto says.

Generally, AI adoption rates for companies’ in-house lawyers have more than doubled in a year, with almost two-thirds of teams expecting to rely less on outside counsel as they strengthen their AI capabilities, according to a report in October 2025 from the Association of Corporate Counsel and litigation software platform Everlaw.

Budgets, however, are not keeping pace. Less than half of legal departments this year expect their internal spending to rise, compared with almost two-thirds last year, according to Cloc (Corporate Legal Operations Consortium), a network of experts, and consultant Harbor.

“They don’t necessarily have the flexibility to throw money at shiny new objects,” says Oyango Snell, Cloc chief executive. “We’re expected to do ‘more with less’. I hate the adage — but the reality is that’s the way leaders are looking at the business. How can we be more efficient? How can we be more effective?”

Leaning on a single AI tool to achieve multiple goals may allow teams to get “two for one, or three for one,” he says.

“One caveat, though, is that while there may have been validation for one use, it doesn’t mean there’s a validation for another use, so you basically increase inherent risk.”

Legal teams face concerns over inaccuracy and must separate hype from reality as they seek to tailor AI tools to solve their specific challenges, Snell adds.

They are still turning to outside specialists in the meantime. AI start-up Harvey is working with more than 500 in-house teams. Legora has added companies including Barclays to its customer list, while teaming up with Ironclad, an AI contract platform, to help in-house lawyers. LexisNexis and Thomson Reuters have strengthened their offerings, partly through partnerships and deals.

But company lawyers are not waiting for products to hit the market. Instead, they are seeking new ways to use AI and investigating an array of options, says Chris Audet, an analyst at research and advisory firm Gartner. One team, he says, is using an HR management platform to track employee compliance in disclosing conflicts of interest, and another is looking to merge its contracts platform with its third-party risk tool.

“They don’t know what is going to deliver the most promising pay-off,” he says. “It’s like being at a buffet. You don’t know what’s good yet. You’re going to come back for seconds. But right now you’re just trying things out.”