A solo lawyer can run a serious AI stack for about $100 a month, and it replaces work that used to require hiring someone. Claude Pro for drafting, an AI notetaker for consults, missed-call text-back so leads stop dying while you are in court, and dictation. Skip the big-firm platforms. Harvey, CoCounsel, and Luminance are priced for firms with research departments.
I started as a solo. One person, no staff, a phone that rang while I was standing in front of a judge. I now own 10 law firms and 62 people work at the biggest one. The thing I would tell my solo self is not about legal research.
It’s this: your bottleneck was never legal knowledge. You know the law. Your bottleneck is that one human being has to do intake, drafting, marketing, billing, and the actual practice of law, and only one of those five bills.
Which means AI is a staffing solution before it is a research tool. Every solo I know who framed it that way got value in week one. Every solo who treated it as a smarter Westlaw quit in three weeks.
The stack
| Tool | Cost/month | What it replaces |
|---|---|---|
| Claude Pro | $20 | Associate drafting hours |
| ChatGPT Plus (optional) | $20 | Second opinion, custom GPTs |
| AI notetaker (Fathom free tier works) | $0 to $20 | Your own post-consult writeup hour |
| Missed-call text-back / basic AI intake | $30 to $100 | A receptionist at $3,000 to $4,000 |
| Dictation (Whisper-based apps) | $12 to $15 | Typing, or a transcriptionist |
| Canva or an AI image tool | $15 | A designer for social and ads |
Add every line and you are over $100. That is deliberate. Two of those rows are optional and you should treat them that way. Claude Pro plus intake plus dictation is the real core, and that lands near $70. Add the notetaker and you are at $90. ChatGPT and Canva come later, if at all.
Start with Claude. The writing holds up better for legal work and it reads long documents without falling apart, which matters when you are the only person who can review them. The full case for it is in Claude for lawyers.
What each line actually buys back
The drafting line is the easiest to price. If a demand letter took you 90 minutes and now takes 20 minutes of review, that is not a productivity tip, that is an associate you did not hire.
The intake line is the one solos underestimate. A receptionist costs $3,000 to $4,000 a month plus payroll tax, and she goes home at 5. Missed-call text-back costs $30 and does not. At my firms the pattern was identical every time we measured it: the calls we lost were not late-night calls, they were 2:15 p.m. calls while a lawyer was in a hearing. A caller who reaches voicemail calls the next firm on the list within four minutes.
The notetaker buys back the hour after every consult where you write down what just happened. The dictation line buys back typing, which is the least valuable thing a lawyer does with his hands.
The five moves, in order
- Cover your after-hours and in-court intake first. Missed-call text-back at minimum, an AI intake agent if your volume justifies it. Do this before anything else. It’s the only item on the list that makes money instead of saving time. The setup detail is in AI intake for law firms.
- Build a prompt library for your five most repeated documents. Not fifty. Five. Your engagement letter, your standard demand, your client update email, your most common motion, and whatever you personally hate writing. Store them where you can paste them in two seconds, because a prompt you have to hunt for is a prompt you stop using.
- Rewrite your client update emails in plain English. Paste the legal version, ask for a version an intelligent non-lawyer understands, read it, send it. Client complaints drop. So do the “any update?” calls that eat your afternoon.
- Automate your consult notes. Notetaker on every call, then a prompt that turns the transcript into a case summary in your format. The summary is written before you have walked back to your desk.
- Write the marketing content you never had time for. One post a week, drafted from something that actually happened in your practice this week. Solos lose to bigger firms on visibility, not on skill, and one post a week beats the polished content plan you will never execute.
What not to buy
Do not buy enterprise legal research AI in year one. The demo is impressive and the price assumes a firm with associates billing under you. Do not buy a per-seat practice platform before you have a workflow it can support, because you will end up shaping your practice around software you barely use. And do not sign an annual contract for anything in your first year with AI. Month to month, always. You will change your mind about half of this stack within six months, and you should.
The real time cost
Two to three hours to set the whole thing up. Then it gives hours back every week.
The failure mode is not technical. It’s the lawyer who opens Claude, types “write me a motion to compel,” gets something generic, and decides AI is overhyped. Of course it was generic. You gave it nothing. The lawyers who win with this treat the first two weeks as training a new hire: more context, corrections, examples of your good work.
If you only do one thing
Turn on missed-call text-back this week. Not the drafting tools, not the prompt library. The intake coverage. It’s $30, it takes twenty minutes, and it starts paying the day you install it because your phone is already ringing while you are unable to answer it.
Then work through the rest in the order above, or follow the full sequence in the AI roadmap.
Today: pull up your call log and count how many calls you missed in the last 30 days. That number is your business case, and it’s usually worse than you think.