Beyond Adoption: AI and the Business of Practicing Law

Adoption is no longer the hard part. Running it well, governing it, and explaining it to clients is.

Ask a managing partner whether the firm is using AI, and the answer is usually yes. Ask which tools, who approved them, and what client information has gone into them, and the answer may take a little longer.

That’s where many firms find themselves. The debate over whether to adopt AI has largely given way to the work of running it. The Thomson Reuters Institute found that organization-wide AI use in professional services nearly doubled in a year, reaching 40% in 2026, and most individual professionals now use generative AI.

The firms that come out ahead won’t be the ones that adopted first. They’ll be the ones that can run it well.

Clients Have Already Set the Pace

Clients are forming their own expectations. In Thomson Reuters’ 2026 Future of Professionals research, 78% of corporate clients said AI-enabled quality improvements are very important or essential. Only 6% said most of their providers deliver them. Nearly a third expect to reconsider provider relationships within the next year because of it.

Pricing is part of that discussion, too. Thomson Reuters’ Raghu Ramanathan notes that roughly 71% of in-house legal professionals expect their outside firms to change their commercial models as AI use grows, while only 28% of firms say they’ve adjusted pricing.

For years, watching and waiting was a reasonable way to adopt technology. Let other firms find the problems, then invest once the tool matured. With AI, clients are the ones keeping score, and they’re not waiting.

Faster Doesn’t Mean Cheaper

AI can shorten parts of an assignment without lowering the overall cost of the matter. BTI Consulting found that 64% of U.S. companies planned to increase litigation budgets this year, driven by complexity rather than volume. Matters are becoming more cross-border, more technology-driven, and more closely tied to business risk.

AI does compress the routine parts: chronologies, document review, issue clustering. But the effort doesn’t disappear. It moves toward strategy, pressure-testing a case theory, and trial preparation. And once one firm can synthesize thousands of documents overnight, everyone else has to keep up.

The result is often more strategically intensive work rather than cheaper work. That puts pressure on the hourly model and brings a familiar leadership question into sharper focus: how does the firm explain the value clients are paying for?

Policy Is Trailing Practice

About a third of professionals say they use AI tools their organization hasn’t approved. Training and formal policy tend to lag behind individual use.

That can become visible in an ordinary deadline. Someone pastes a draft into a free tool to tighten the language. They’re trying to finish the work, and nothing approved was easier to reach.

A ban alone doesn’t necessarily stop that use. It can move into personal accounts and unapproved applications where the firm has little visibility and no practical way to audit it. Approved tools, clear boundaries around submitted information, and defined verification steps give people a workable process within the firm.

Some firms carry that further by marking each matter record as AI permitted, prohibited, or conditional. The client’s outside counsel guidelines then become part of the workflow, available when an attorney is working against a deadline at six in the evening.

The instruction is there with the matter. Its application depends less on someone remembering a policy at the end of a busy day.

Confidentiality Shapes Every Decision

Law firms hold privileged communications, financial records, and personal information together. That concentration makes them an attractive target, and it shapes how AI fits into the practice.

Client requirements add another layer. Some outside counsel guidelines prohibit confidential information from entering certain AI systems altogether. Others require disclosure.

Open platforms may use submitted information for training, depending on the service and account terms. Closed platforms are built with safeguards intended to limit exposure, although those safeguards don’t remove every consideration. Removing names, Social Security numbers, and other sensitive details wherever possible remains useful even within a closed system.

One law firm administrator recommended developing AI efforts in close collaboration with IT and security leaders from the beginning. That puts information handling, access, and protection into the discussion while the application is being considered.

When the tool is chosen first, those questions arrive later, sometimes after people have already begun using it. Addressing them earlier gives the firm a clearer basis for adoption. Phishing-resistant multi-factor authentication and consistent everyday security hygiene remain central to protecting the firm. Safe AI use rests on that same foundation.

Someone Has to Stay in the Driver’s Seat

Most attorneys are familiar with the warnings about AI-generated errors and fabricated citations. Sanctions have made the consequences clear. The more useful discussion now is how that awareness carries into the firm’s daily work, especially as AI moves beyond drafting answers to taking actions.

These systems are probabilistic, so reviewing an answer doesn’t provide the same visibility as questioning a junior attorney about their research and reasoning. The firm can examine the sources and test the conclusions, but the process that produced them is less transparent.

The tolerance for uncertainty also depends on the assignment. A marketing draft allows room for variation. A court filing or privileged client advice carries different consequences. The duties of competence, confidentiality, and supervision remain in place, and work issued under the firm’s name remains the firm’s responsibility.

Robert Christensen, writing for the ABA, framed the broader question well: are professionals driving their systems, or riding along?

That question becomes more consequential when an AI agent can edit records, save changes, or send correspondence. Its permissions, the environment it operates within, and the points where a person reviews and approves its actions become part of how the firm maintains control.

For clients, that responsibility comes back to four plain questions: what do you use, why do you use it, how do you protect my information, and what difference does it make to the service I receive?

What AI Depends On

AI becomes part of the firm’s existing environment. Information access, managed devices, and security continue to shape how dependably it can be used. Clear permissions, current systems, timely removal of access when someone leaves, and visibility into unusual activity all contribute to that foundation.

At Handled IT Partners, we look at how those pieces work together, what’s already serving the firm well, and where improvements would support its plans for growth. From there, we help evaluate and implement services and technologies that fit the practice, giving leadership a clearer basis for decisions about AI and its use with client information.

If AI is on your agenda, we’d welcome a conversation. Your clients’ outside counsel guidelines, alongside the applications your team uses today, offer a practical place to start.

Schedule a quick 15-minute call with our team.