AI for Solo Practitioners: The Setup Before the Tools

For most solo lawyers in 2026, AI is a general-purpose chat tool used for first drafts and summaries, with no written rule about what goes into it and no change to how the work is priced. Clio's 2026 solo and small firm report found that 71% of solo practitioners use AI and 57% have no AI policy. The gap that matters is not which tool you pick. It is the handful of decisions nobody has written down.

This post is for two kinds of readers. The solo lawyer who already has a chat window open and wants to use it well, and the two-to-five-lawyer firm that is starting to wonder whether something should be built. Nothing here is legal or ethics advice. Your state bar's guidance and your tool's data terms win every time.

What does AI for a solo practitioner actually look like in 2026?

The numbers from two surveys line up.

Clio's 2026 Legal Trends Report for Solo and Small Law Firms, as summarized by the North Carolina Bar Association on May 20, 2026, reports:

• 71% of solo practitioners and 75% of small firms have adopted AI.
• 57% of solos and 55% of small firms have no AI policy.
• 86% of solo firms and 78% of small firms have not adjusted their pricing to account for AI.
• About a third report a revenue increase from AI: 32% of solos and 31% of small firms.
• For 27% of solos, finding time is the biggest hurdle to adoption.

8am's 2026 Legal Industry Report surveyed more than 1,300 legal professionals between September 19 and October 18, 2025. Solo practitioners were 45% of respondents and firms of two to five lawyers another 38%, so it is mostly a picture of small practices. As LawSites reported on March 5, 2026, 69% of respondents now use general-purpose AI tools for work, up from 31% a year earlier.

Why is 71% adoption not the same as a practice that uses AI well?

Put the 8am numbers side by side:

Individuals ran ahead of their firms, which makes sense. Opening a tab takes ten seconds. Deciding a rule takes an afternoon and a little discomfort.

For a solo, "the firm" and "the individual" are the same person, which is both the problem and the advantage. There is no committee to wait on. There is also nobody to catch the day you paste a client's full name and a privileged fact into a consumer chat account at 11 p.m.

Both the risk and the upside sit in the undecided part. Confidentiality is the reason the decisions come before the tools. How your jurisdiction applies the confidentiality and competence rules to AI is your bar's call, so read their guidance rather than a vendor's summary of it.

What are the three decisions to write down before choosing a tool?

This is the policy 57% of solos do not have. It fits on one page. We would write it as three short sections.

Decision one: what never goes in

Be concrete. "Be careful with client data" is not a rule. A rule looks like this:

• Which account. A business or team plan with data terms you have actually read, not a personal consumer login. Check whether your inputs are used for model training and whether you can turn that off.
• What gets redacted by habit. Client names, matter numbers, opposing parties, and privileged facts come out before anything is pasted, unless the tool's terms and your bar's guidance cover putting them in.
• What never goes in at all. Pick the category for your practice. For some solos that is anything under seal. For others it is a whole practice area.

Decision two: which tasks get an AI first pass

Write down the short list of jobs where a draft from a chat tool saves real time and a mistake is easy to catch:

• Summarizing a long file or a deposition transcript you will read anyway.
• Distilling a 40-message email chain into what was agreed and what is still open.
• A first draft of a routine letter or a standard clause you will rewrite.
• A plain-English explainer of a process for a client, checked by you before it goes out.

Then write down what is off the list. Filings are off. Legal research you have not verified against the source is off. The list is short on purpose.

Decision three: who reviews, and how

The lawyer reviews every output before it leaves the office. Every one. Decide the habit that makes that real, for example a rule that nothing AI-drafted is sent until you have compared it against the source, and a quick note in the file of what the tool drafted and what you changed.

We built a clinical notes tool for dentists on exactly this rule: the AI drafts the note, and the dentist reviews and signs every one. You can read how that shipped in the Smile PreVue case study. The profession is different. The principle is the same. The drafter is never the signer.

How should a solo pick between a general tool and a legal-specific one?

Frame it by job, not by vendor scorecard.

General-purpose tools (Anthropic's Claude, OpenAI's ChatGPT, Google's Gemini) are strong at the three tasks above and cheap relative to what they save. For drafting, summarizing, and explaining, any of the three can do the work. The differences that matter for a lawyer are the data terms on the plan you buy and how comfortable you are working with it every day, not a benchmark chart. We use all three labs across our own builds, and we pick per problem.

Legal-specific tools earn their price when they connect to something you already pay for, like your practice management system or a research library, so the AI works inside your matter files or cites a source you can check. If a legal tool does not connect to anything you use, you are mostly paying for a different chat window.

If you want to get genuinely good at a general tool, not just open it, that is what our 1:1 Claude training is for. We sit with you and your real work.

Why do the hours saved not show up as revenue?

This is the part of the Clio data that surprises people. Most solos report using AI. Only about a third report more revenue from it. And 86% of solo firms have not changed how they price.

The math explains it. 8am found that 38% of individual users save one to five hours a week and 14% save six to ten. Under hourly billing, a task that used to take five hours and now takes one is four hours you did not bill. You handed the client a discount without deciding to.

There are several ways firms respond: flat fees for defined work, a different mix of matters, more matters in the same week, or simply more time back. Which one fits your practice, your clients, and your bar's rules on fees is your call, not ours. The operator point is only this. If you have not decided, the default is the discount.

When should a small firm build instead of buy?

A solo with no repeatable volume should not build anything. Use a general tool well, write the one-page policy, and stop there.

A two-to-five-lawyer firm is different when it has a repeatable pipe: intake forms that arrive the same way every week, conflict checks that always pull the same fields, or a first pass on the same kind of document fifty times a quarter. That is where AI integration for business starts to pay back, because the work is predictable enough to design around.

Our method for that kind of build is the same one we use everywhere: Figure it out, Build it, Ship it.

• Figure it out. Sit with the people doing intake and find the one step that eats the most time and has the clearest right answer.
• Build it. Build only that step, with the lawyer's review built into the flow, not bolted on after.
• Ship it. Put it in front of real matters, measure what changed, and decide the next step from what actually happened.

The review step is not optional in a law firm build. It is the design. Our earlier post on AI for law firms, what actually ships and what stays a pilot goes further on sequencing.

Frequently asked questions

Is it safe to put client information into Claude or ChatGPT? It depends on the account, the tool's data terms, and your bar's guidance, which is why decision one comes first. A business plan with training turned off is a different situation from a personal login. Never assume a blanket yes.

Which AI tasks are safest for a solo lawyer to start with? Summaries of material you will read anyway, email-chain digests, first drafts of routine letters, and plain-English client explainers. All of them are easy to check against the source.

Do I need an AI policy if I am the only lawyer? Yes, and it is one page. 57% of solos do not have one, according to Clio's 2026 report. The page is for you on a tired Thursday night, not for an auditor.

Should I switch to flat fees because AI made me faster? That is your decision, and your bar's fee rules apply. The data only says this: most solos have not changed pricing, and only about a third report revenue growth from AI.

Start with the page, not the tool

Write the three decisions down this week. What never goes in, which tasks get a first pass, and who reviews. That one page puts you ahead of most of the profession.

If you want to learn to use Claude well on your own work, start at /learn. If your firm has a repeatable intake or document pipe and wants it built with the review step designed in, tell us about it.