AI for law firms
AI for law firms that leaves the legal judgment with lawyers.
AI is most useful to a law firm before a lawyer sees a matter: reading every inquiry, pulling out the facts, and routing each one to the right person on a clock. It should never decide whether a case has merit. I build intake systems where the model cannot make that call, on accounts your firm owns.
By Jimmy Hackett, founder of Playground Giants · Updated September 27, 2026
Where AI helps a law firm, and where it has to stop
The model does this
- Reads inquiries from forms, email and calls
- Extracts the facts: what happened, when, where, who else was involved
- Hands off to a person whenever it is unsure or unavailable
Your firm does this
- Writes the rules: limitation periods, thresholds, conflicts
- Decides whether to take a matter, always a person
- Judges merit and value, which no grade ever describes
- Owns the accounts, the data and the audit trail
The split is in the architecture, not in a policy document. Ordinary code applies your rules to the extracted facts, so the same facts always get the same grade, and a lawyer can read the rule and change it.
Legal intake services, built into your own systems
Most legal intake services put people between you and the client. This is the layer underneath: whoever answers, every inquiry lands in one place, graded on its facts and on a clock.
One intake path
Website inquiries and calls captured into a single pipeline, so nothing depends on who happened to pick up.
Grading on facts, never merit
Each inquiry is graded against rules your firm writes. The grade sorts the queue. It is not an opinion about the case.
Response clocks that escalate
Urgent matters carry a thirty minute clock, standard matters a day. A missed clock escalates to a named person instead of expiring quietly.
Serious categories skip the queue
Matters you name route straight to a person, whatever the grade says.
Legal intake solutions you own, not rent
An off-the-shelf legal intake solution keeps your intake history on its vendor's platform. A built one lives on accounts in your firm's name.
An append-only audit trail
Kept for ten years: what arrived, what was extracted, which rule applied, who acted, and when. Written once, never edited.
Fail toward a person
If the model is down or unsure, the inquiry goes to a human. An outage cannot silently drop a matter.
Checked every day
A daily digest lists every inquiry where the AI's grade and a person's grade disagreed, so the rules get fixed rather than argued about.
Everything in the firm's name
Hosting, database, analytics and ad accounts. If you replace me, you change the passwords and keep operating.
Running today
A New York personal injury firm
This is how intake runs for a New York personal injury firm, which is not named at its preference. Since September 21, 2026, new inquiries have been graded on their extracted facts as they arrive, and a person still reviews the top grade. The grader was tested against 66 made-up inquiries written to the firm's own rules, and a daily digest has flagged every disagreement between the AI and a person since September 24.
There are no outcome numbers yet. The engagement has not reached its first formal measurement point, and any cost-per-case figure printed now would be a projection.
Read the full case studyHow it starts
With a two week Workflow Audit. I map how inquiries reach your firm today, mark what is safe to automate and what has to stay with a person, and price the build in phases. It is a $2,500 fixed fee, credited in full if you build within ninety days, and the plan is yours either way.
The stages after that are on the packages page.
Questions law firms ask
What can AI actually do for a law firm?
The safe, useful work sits before and around the legal judgment: reading inquiries that arrive by form, email and phone, pulling out the facts, checking them against rules the firm writes, routing each matter to the right person on a response clock, and keeping a record of every step. It should not decide whether a case has merit or what it is worth. That stays with your attorneys.
Does the AI decide which cases we take?
No, and the system is built so it cannot. The model reads a message, extracts facts, and stops. Ordinary code applies your firm's own rules to those facts, and a person makes the call. Anything the model is unsure about goes to a person rather than being guessed at.
What is the difference between legal intake services and legal intake solutions?
Usually, a service is people answering and screening for you, and a solution is software you use yourself. What I build is the second kind, on accounts in your firm's name: one intake path, grading on facts, response clocks, escalation and an audit trail. Your own staff, or an answering service you already use, still talk to the client.
How does this stay inside attorney advertising and ethics rules?
By keeping the software away from anything that counts as legal judgment. Grades describe facts, never the merit or value of a claim. A person reviews the top grade before anyone is contacted. Every step is written to an append-only log that you can show a regulator or an insurer. Where a question needs your own counsel's sign-off, it is part of the scope.
Do we own it?
Yes. Hosting, database, analytics and ad accounts are set up in the firm's name. If you replace me, you change the passwords and keep operating.
Do you need access to client files?
Not to scope the work. A conversation and screenshots of how inquiries arrive today are usually enough to map the workflow. I never ask for privileged material to price a build.
What does it cost to start?
Nearly every engagement starts with the Workflow Audit: two weeks, a $2,500 fixed fee, and a written plan that is yours either way. If you go ahead with a build within ninety days, the full fee comes off the build price.
Find out what your intake could hand off.
Fifteen minutes is enough for me to tell you whether there is anything worth building. If there is not, I will say so.