MMMRMossPro
← Field Notes  · mrmosspro.com
Field Notes · Nº 5 · AI-Enablement for Legal

What I discovered for solo practitioners

The whole point, up front
What I discovered for solo practitioners — while building the AI, not billing an hour:
  • A paid research tool may already be free in your bar dues. Mine was. I almost bought it anyway.
  • An AI can do the tedious verification with youthe associate’s hours a solo never had.
  • The one check that decides everything stays human. The machine flags; you rule.

I don’t run a law practice. My work is building the AI operating systems that run other people’s — AI-enablement for legal, full stop. So when I put one of these systems through real legal workflows, I’m not lawyering; I’m pressure-testing what the solo on the receiving end will actually experience. This week the test handed me three things I’d give any solo practitioner in a heartbeat — and the first had nothing to do with clever prompting and everything to do with what a solo already owns. Notes sixteen through eighteen.

16. Solos do have a friend in the AI

The system needed to confirm that court decisions were still good law — that nothing later had overruled or undercut them — which takes a citator. Citators are the kind of paid tool a big firm buys site-wide and a solo squints at the price of. So I did the solo thing: I started filling out a “contact sales” form to go buy one. The AI stopped me mid-form. Wrong door. The access was already there — a state bar had folded a citator (Decisis) in as a member benefit, free to every active licensee through the practice portal, and it had simply never been switched on. Not a trial, not a callback. A login already paid for, gathering dust.

That is the discovery I want in front of every solo: check what your bar already gives you before you reach for a credit card. And then the second half, which is the part money can’t shortcut — the AI ran the manual checks alongside, one authority at a time, surfacing each result so a human could read it. The gap between a solo and a large firm was never only the subscriptions; it’s the associate who knows which benefit is buried in the dues, and the hours to actually run the tool once it’s found. An AI-enablement system can be both — the colleague who reads the fine print and the second set of hands on the tedious pass. That is not a small thing when the whole firm is one person.

17. A red flag is a lead, not a verdict

Partway through those checks, a brand-new opinion lit up with the worst thing a citator can say: negative treatment detected. The reflex — the one an unattended script would follow — is to drop the case and move on. But a flag is a place to start reading, not a verdict to obey. Open the actual report and the “negative” items weren’t about the case at all: they were unrelated decisions that merely shared a common surname, parked in a sidebar of related documents. The opinion itself was clean.

For a solo evaluating any of these tools, that’s the pattern worth internalizing: automation is fastest at producing a signal, and slowest at knowing what the signal means. A scraper obeys the badge and quietly kills a perfectly good authority; a practitioner reads two more screens and keeps it. The value was never the automated flag. It was refusing to treat the flag as the answer — which is exactly the kind of judgment a good enablement system is built to route to you, not around you.

18. The check that matters most is the one a machine can’t run

Two things keep that citator pass in human hands, and they point the same way. First, the seat is an individual member benefit — its terms forbid scripting and bulk automation, so the lookups genuinely can’t be mechanized like the deterministic gates. Second, as the false alarm showed, reading a flag correctly takes judgment a script doesn’t have. So the most consequential verification in the whole run is, by design, the one a human drives.

There’s an honest tail I build for openly. Because that check is manual, the automated battery can go green while it’s still unrun — every deterministic gate passes, and the one that needs a person hasn’t happened yet. A green board is not a finished job when a manual gate is outstanding. So the discipline I ship to a solo is boring and load-bearing: a manual gate that lives only in someone’s head is a gate that eventually gets skipped, so it goes on the checklist beside the automated ones, with the same authority to stop a shipment. The system does the labor with you. It does not do the judgment for you — and an honest one is built so it can’t pretend to.


The takeaway — for the solo, not the firm. “Verification is the product” has a second meaning when you’re one person: it’s the verification you couldn’t afford, or didn’t know to run, or never had the hours for. A well-built enablement system closes all three — it finds the benefit already sitting in your dues, does the manual pass beside you instead of leaving it as homework, and at the decisive step reads the flag with you and hands the call back. Not a replacement for the practitioner. The associate, the librarian, and the second set of hands a solo has never had — with the last judgment kept firmly human.

These notes aren’t a practice diary — they’re a record of what I keep finding while building AI systems for legal work, written for the solo and small-firm attorneys those systems are for. This one you can act on without me: go log in to your bar’s practice portal and see what’s already yours.

Field Notes are written for licensed attorneys and are not legal advice. Mike Moss is a Utah-admitted attorney doing AI-enablement work — not operating a law practice.