AI in Law Firms: Getting Ready for Clients Who Use AI Too

At a Glance

• Who this is for: Heads of IT, innovation leads and COOs at UK firms whose clients now run AI of their own.
• The problem: Firms have spent two years deciding how to use AI. Their clients have done the same, and now arrive with their own tools, their own drafts and their own cost targets.
• What is changing: 85% of senior in-house legal leaders expect AI to change how law firms price, and they expect hourly-rate work to fall from 72% to 44% within two to three years.
• The takeaway: Every question an AI-literate client asks about your AI ends in a request for evidence, and evidence is something your systems either produce or don't.

AI in law firms used to be a question about the firm: which tools, which matters, what the insurer would say. That question now has a second half, because the client on the other side of the engagement letter is running AI too.

A general counsel can paste your draft into their own model before your call starts. A business owner can arrive with a contract a chatbot wrote and ask you to check it. Both expect something different from the firm than they did two years ago, and most of what they expect is technical.

AI in Law Firms Now Has a Client Side

Deloitte Legal surveyed 121 senior in-house legal leaders between April and May 2026. Just 2% reported no AI adoption in their department, against 76% in 2024, and 61% are already experimenting with or piloting agentic AI (Deloitte UK).

The same leaders expect AI to save or automate an average of 28% of their legal work over the next two to three years. Some of that work is billed by outside counsel today.

The consumer side has moved as well. A Legal Services Board survey of 1,000 adults in England and Wales, published in June 2026, found 70% expect AI to make legal services easier to use and 64% expect it to make them more affordable. People already use AI tools on housing, employment, debt and family problems.

So there are now two kinds of AI-literate client, and each creates different work for the firm.

What In-House Legal Teams Are Doing With AI

In-house teams are using AI to bring work back in-house, give the business self-service answers and run legal operations with fewer people. What reaches the panel firm is the work AI could not absorb, plus a cost target.

Legal Futures' report of the Deloitte findings notes that some general counsel are setting cost reduction targets of 20 to 40% over two to three years and passing them on to outside counsel. 78% of the leaders surveyed named cost reduction as the benefit they most want from their firms' use of AI.

The awkward number for firms is a different one. 58% said their external firms rarely or never raise the benefits of AI with them, and only 4% had directly seen a benefit from a provider's AI.

We hear the same thing from the firm side. Across our meetings with UK firms in September 2026, from a top-25 practice to a regional firm with aggressive growth targets, the commercial brief was consistent: clients want lower rates, so firms are using agentic workflows to take cost out and pass some of it on. The firms further along could show where the saving came from. The rest could only say it was coming.

Clients Who Arrive With an AI Draft

The second group are individuals and smaller businesses who now do a first pass with a general-purpose AI tool before they call a solicitor. They arrive with a draft, a list of clauses they want changed, or an answer they want confirmed.

That changes the intake job. A client-supplied draft may cite law that doesn't exist or apply English law to a Scottish problem, and it carries no record of what was asked to produce it. The SRA's Misuse of AI warning notice is clear that the solicitor who relies on an authority is responsible for checking it, wherever the text came from.

Consumers also told the LSB what they won't trade for convenience: a minimum accuracy guarantee, informed consent before any consequential action, human oversight at all times, redress for harm, and protection of their safety and personal information. That is a specification a firm can build to.

Comfort with firms' own AI use is mixed. In a Law Firm Marketing Club survey of 642 clients, reported by Legal Futures in April 2026, 45% were comfortable with AI supporting the delivery of legal services and 32% were uncomfortable. Clients wanted reassurance that a named lawyer stays responsible for their matter.

Where the Pressure Lands: Price, Panels and Proof

The billable hour takes the first hit. 85% of Deloitte's respondents believe AI will change how firms price, and they expect hourly-rate work to fall from 72% of the work to 44%. Fixed fees and other alternative fee arrangements put the efficiency gain where the client can see it, but only if the firm knows its own cost to deliver a matter. We covered the commercial side in how top law firms are rethinking pricing.

Panels come next. When a legal department reviews its panel, AI questions now sit beside rates, and some are starting to appear in outside counsel guidelines. The questions are technical even when the person asking isn't: where client data is processed, whether it trains a model, who reviewed the output, and how time saved shows in the bill.

Then proof. A client that runs its own AI can test yours. If a GC can summarise a disclosure bundle in minutes with their own tool, a long time entry for the same summary will get queried, and the firm needs to show what those hours covered.

Five Answers Your Firm Should Have Ready Before a Client Asks

Each answer below depends on something the firm's systems record. If the record doesn't exist, the answer is a partner's recollection, and an AI-literate client will notice the difference.

Rows 2 and 3 are where most firms come up short. A tool register is a spreadsheet; a review log has to be built into the drafting or document management workflow, and it's the record a client or a regulator will ask to see. We set out how to build it in AI governance for law firms.

Row 2 gets much shorter when the AI runs inside the firm's own Microsoft tenant. The answer to "where does our data go" becomes "nowhere it wasn't already", which is the answer most GCs are hoping to hear.

If nobody at your firm owns these answers yet, start with who should own AI at a law firm.

How 3Rive Helps

Our AI services team builds legal AI tooling that runs inside the firm's own tenant, including court-ready redaction, knowledge mining and matter chronologies, so client data stays in the firm's environment and the review step is part of the build. Through Tech Advisory, AI Readiness in Five Days maps each tool's data flow, names a reviewer for each type of AI-assisted work and ends with a costed 90-day plan. We set out what that week covers in AI readiness assessment for law firms.

Key Takeaways

• Just 2% of senior in-house legal leaders reported no AI adoption in 2026, against 76% in 2024.
• 85% expect AI to change how law firms price, and 58% say their firms rarely or never raise AI's benefits with them.
• Consumers expect AI to make legal services easier and cheaper, and still want accuracy guarantees, human oversight and data protection kept in place.
• Client questions about a firm's AI end in a request for evidence: a tool register, a data-flow record, a review log and a pricing position.
• A solicitor stays responsible for checking any authority they rely on, including one in a draft the client produced with AI.