Here is a strange result from this year's legal surveys. Small law firms have adopted AI faster than almost anyone expected. Lawyers say it saves them hours every week. And most of those firms are not making any more money. If you run a small practice, that is not a technology problem. It is a pricing problem.
What happened
Clio's 2026 report on solo and small firms, summarized by the North Carolina Bar Association in May, found that 71% of solo practitioners and 75% of small firms use AI. But only 32% of solos and 31% of small firms report a revenue increase tied to it. Meanwhile, 86% of solos and 78% of small firms have not changed their pricing to account for AI. The article makes the key point plainly: for hourly billers, time saved works like an unplanned discount unless it is filled with other work. It also found that more than half of solos and small firms have no AI policy.
The 8am 2026 Legal Industry Report, covered by LawSites in March, shows the same pattern from a different angle. Sixty-nine percent of legal professionals now use generative AI for work, up from 31% a year earlier. Among users, 38% save one to five hours a week. Yet 43% say their firm has no AI policy and no plans for one, and only 9% have a written policy that is actually enforced. About a quarter of those expecting billing changes anticipate fewer billable hours per matter. And clients are not pushing yet: only 6% of respondents report clients asking for AI-related price cuts.
The risk side has not gone away either. The ABA Journal reported on October 1 that researcher Damien Charlotin's database now counts 1,430 U.S. court cases involving AI hallucinations, and that two lawyers received a record $110,000 fine this year.
Why it matters for your business
Do the math on a small firm that bills by the hour. An associate uses AI to draft a first-pass motion in two hours instead of five. If the firm bills two hours, it just earned less on that task. If it bills five, it has an ethics problem. Neither is a business model.
That is why the revenue numbers lag. The firms seeing gains are the ones that changed how they charge, not just how they work. Clients are not demanding discounts today, which gives small firms a window to set their own terms before the market sets them.
The policy gap matters to the money too. A firm with no rules about which tools to use, what client data goes where, and who checks the output is carrying risk it is not pricing in. One sanctioned filing can cost more than a year of saved hours.
The defenders of hourly billing have a point: it is transparent, clients understand it, and many matters are too unpredictable to price as a flat fee. Fair enough. But routine work like simple wills, standard leases, demand letters and uncontested filings is predictable, and that is exactly the work AI speeds up most.
What we'd do
- List your five most repeated matter types. Track how long each takes now with AI help, not how long it used to take.
- Move the predictable ones to flat fees. Price on the value to the client, informed by your old hourly totals, so the time saved becomes margin.
- Fill the freed hours on purpose. More matters per lawyer, faster turnaround, or time for business development. Decide which, or the hours will simply disappear.
- Write a one-page AI policy. Approved tools, client data rules, and a required human check on every citation and quote before filing.
- Use legal-specific tools for research. General chatbots are fine for drafting and summarizing your own documents, not for finding case law.
- Measure quarterly. Revenue per lawyer and realization rate tell you whether AI is paying, not hours logged.
Our view: small firms have done the hard part by adopting AI. Now they are giving the benefit away. The billable hour was built for a world where legal work took a fixed amount of human time. That world is ending for routine matters, and the firms that reprice first will keep the savings. The rest will hand them to clients who never asked for them.
Want a system like this in your business?
A discovery conversation is where we map it: what to automate first, where the human checkpoints go, and what it would take.
Book a discovery conversation →