AI for Lawyers: Where the Billable Time Goes

The economics of a small law firm are straightforward: billable hours are revenue, everything else is overhead. The problem is that a significant chunk of every week goes to work that doesn't bill: intake, document drafts, client follow-up, scheduling, status updates, file management. Work that has to happen but doesn't pay.

That's where AI fits in a law firm. Not in the legal work itself. In everything around it.

Where AI actually helps

Client intake

Most firms collect the same information from every new client, just in a slightly different order each time. An intake form that covers everything upfront, connected to an AI summary that pulls out the key facts and flags what's missing, cuts the back-and-forth that currently takes two or three days. The attorney walks into the first meeting with a clean summary instead of a pile of emails.

The risk here is low. The AI isn't practicing law. It's organizing information the client already provided.

First-draft documents

Engagement letters, retainer agreements, standard demand letters, NDAs, straightforward contracts: documents with predictable structure that get rewritten from scratch or pulled from a file somewhere and manually edited every time. AI produces a solid first draft from a brief. You review it, edit it, sign off on it. The blank page problem goes away.

This works best for documents where the structure and standard language are consistent and the variation is in the specific facts. It works less well for complex or novel legal questions where the drafting itself is the judgment call.

Meeting notes and follow-up

After a client call, AI turns rough notes or a transcript into a clean summary with open items and next steps. The follow-up email drafts itself from the same summary. Neither is particularly hard to do manually, but together they add up to 30 to 45 minutes per client meeting, and most attorneys have several a week.

Client communication

Status update emails, document request follow-ups, deadline reminders, answers to common questions. These follow patterns. AI handles the draft. The attorney reviews it, adjusts the tone if needed, and sends. Faster than writing from scratch, better than skipping it.

Research summaries

AI is useful for pulling together background on a topic, summarizing a document, or drafting a quick overview of a legal area before going deeper. It is not a replacement for actual legal research. It gets things wrong in ways that look right, which is a serious problem in a legal context. But as a starting point before you open Westlaw, it saves time.

Where to be careful

Any output that goes to a client without review

The workflow is always: AI drafts, attorney reviews, attorney sends. A draft that goes out unread is a malpractice risk, not a time saver. The review step isn't optional.

Legal research and case law

AI confidently cites cases that don't exist. This is well documented and hasn't been fully solved. Use AI to orient yourself on a topic, then verify everything through proper legal research tools before it goes anywhere near a brief or a client.

Confidential client data

Before feeding any client information into an AI tool, know where that data goes and what the tool's data retention policy is. Most of the major enterprise AI tools have terms that address this. Most small firms haven't checked. Check.

Anything that requires judgment specific to the client's situation

AI is good at structure and pattern. It's bad at nuance that only becomes visible after years of practice in a specific area of law. The further you get from standard and repeatable, the less useful AI becomes.

The framing that makes this work

The attorneys who get the most out of AI think of it as a first-pass machine, not an answer machine. It removes the blank page, organizes the information, drafts the routine. The attorney's job is still the judgment: on the law, on the strategy, on the client. AI just clears the path to get there faster.

Where to start

Intake summaries or meeting follow-up emails. Both are low risk, both happen frequently, and both will show a return in the first week. Pick one, build the workflow properly, run it for a month.

If you want to look at the whole practice at once, the AI Workflow Audit is where to start. We find the highest-leverage places in your specific firm and give you a plan you can act on.