home  /  the future of AI in legal
department of the future of AI in legal

The tools are a small change. The business model is not.

Adoption has roughly doubled in a year. The billable hour, which still accounts for most of what firms charge, has no comfortable answer to work that takes a fifth as long.

begin here

Where is your firm?

Start a conversation with the AI Adoption Concierge, already scoped to the future of AI in legal. Choose the question closest to yours, or describe your situation directly.

AI Adoption Conciergethe future of AI in legal · orientation, not legal or ethics advice
Tell me how the firm makes money — practice mix, how matters are priced, and how leveraged the work is. I'll help you work out where the pressure lands first. I won't predict the future, and I'll say when something is genuinely unsettled.

Most firm conversations about the future of AI are conversations about tools, and tools are the least interesting part. The consequential questions are structural: what happens to pricing when the hour and the value of the work come apart; what happens to the associate pyramid when the tasks juniors learned on are the tasks that automate first; and what happens to the firm's position if clients can buy some outcomes directly from software rather than from lawyers. Reported adoption moved from around 28% of firms to around 41% in a single year, with corporate legal departments moving faster still. That pace does not settle any of the structural questions — it only shortens the time available to think about them.

specialization areas

Areas in this part of the practice.

Three structural shifts, each already underway, each with a decision attached for firm leadership.

methodology

How this department investigates.

How the Institute approaches the forward view — observed change and named uncertainty, never prophecy.

Autonomy assessmentHow much a system decides on its own, and where a human must remain in the loop for the work to be defensible.
Pricing pressure analysisWhich matters are exposed first when the hours fall and the value does not.
Pyramid modellingWhat leverage looks like when the bottom of the pyramid is the part that automates.
Disintermediation riskWhich work a client could buy from software instead of from the firm, and how soon.
Capability buildingWhat the firm needs its people to be good at in five years that it is not training for now.
SequencingWhat has to be decided this year rather than watched, which is a shorter list than it appears.
common questions

The future of AI in legal — the questions firms ask.

Is the billable hour actually finished?

No, and the more useful framing is that it is under specific rather than general pressure. Estimates put hourly arrangements at around 80% of law firm fees, and that does not unwind quickly. What is happening is narrower: on matters where AI genuinely compresses the work, a firm billing by the hour bills less for delivering the same result, which is a direct conflict between doing the work well and being paid for it. That pressure lands on high-volume, document-heavy, repeatable matters first, and barely touches bet-the-company litigation. Most firms will run mixed models for years rather than switching.

What is agentic AI and is it real yet?

An agentic system is one that plans and executes a sequence of steps toward a goal rather than answering a single prompt — running a search, reading results, drafting, revising, calling other tools. Several large firms have deployed such systems internally, so it is real in that sense. The version to be sceptical of is the fully autonomous one: the Law Society has flagged that solicitors remain professionally responsible for outputs they cannot fully audit, which is a genuine and unresolved gap rather than a transitional annoyance. Systems that escalate decisions to a human and keep an auditable record are the ones that fit professional obligations as they currently stand.

Will firms need fewer associates?

The honest answer is that nobody knows, and the disagreement is substantive rather than a hedge. One view: AI automates much of what first- and second-year associates do, so firms need fewer of them. The other: demand for legal work expands when it gets cheaper, and firms will need people who can supervise AI output at volume, which is a different skill but still a headcount. What is much less disputed is that junior roles change first and most — less first-draft production, more review, validation and oversight — and that firms training juniors exactly as they did in 2020 are training for a job that is changing underneath them.

Could clients stop needing law firms for some work?

For some categories, this is already happening at the edges, and the trend worth watching is the shift from AI as a tool lawyers use to AI as a service sold directly to the client. Corporate legal departments are adopting faster than firms are, and where an in-house team can get an acceptable answer from software, the work does not reach a firm at all. This affects commodity work first — routine contracts, standard filings, initial research — and it is not obviously a catastrophe, since much of that work was already leaving via alternative providers. It does mean the firm's position rests increasingly on judgement, relationship and accountability rather than on production.

Thinking past the pilot?

Describe how the firm makes money today. The Institute will help you see where the pressure lands first.

AI adoption conciergeorientation · not legal or ethics advice
Tell me how the firm makes money — practice mix, how matters are priced, and how leveraged the work is. I'll help you work out where the pressure lands first. I won't predict the future, and I'll say when something is genuinely unsettled.