Claude for Lawyers: How I Actually Use It in Practice

I get asked which AI tool to use more than any other question, and the honest answer is that the tool matters far less than the discipline around it. That said, I use Claude for the bulk of my own drafting, analysis and system-building work, and I want to describe specifically how, because "I use AI" is a sentence that means almost nothing on its own.

Turning a rough system spec into structured documentation

When I am building a system, the first artifact is always the paper version, matter types, fields, action sequences, written down before any software gets touched. That paper version is often rough: bullet points, half-sentences, the shape of a thing before it has a name. I use Claude to take that rough spec and turn it into structured documentation, a clean field list, a numbered action sequence, a first-pass data model, that I can then hand to whatever tool I am actually building in. This is not Claude designing the system. It is Claude doing the transcription and structuring work fast, so I spend my time on the actual design decisions instead of formatting.

Drafting client-facing explanations of technical structure

After a system is built, someone has to explain it to the client in plain language, what the intake stage does, why the evidence anchoring matters, how the deadline tracking works. I draft the technical structure myself, then use Claude for a first pass at the plain-language explanation, because writing the same explanation for the tenth different client, in a slightly different tone each time, is exactly the kind of repeatable-but-not-identical task AI handles well and I handle slowly.

Long-document analysis before a build

Before building a practice-area system, I often work from a client's existing process documentation, an old procedures manual, a set of email threads describing how intake currently works, sometimes a genuinely disorganized pile of notes. I use Claude to do a first pass through that material and surface the actual process hiding inside it: what happens, in what order, who does each step. I do not trust that first pass as the final word, I verify it against the client directly, but it turns a multi-hour reading task into a twenty-minute review task.

Reviewing my own writing for drift

This site's copy, product descriptions, essays like this one, go through a Claude pass specifically looking for drift: places where the language has slid into generic vendor-speak instead of saying something specific and true. I do this because it is a blind spot every writer has for their own work, and a second pass looking for a specific failure mode catches things a general "does this read well" review does not.

What I do not use it for

I do not use general-purpose AI for anything that requires knowledge of a specific matter's actual documents, because a general assistant only knows what gets pasted in, and pasting in enough context to be useful, every single time, is itself the workflow problem I am usually trying to solve for a client. That is the actual argument for a matter-wired system like MatterOS over an open chat tab: the difference between AI you have to feed context and AI that already has it.

I also do not let it draft anything that goes in front of a client or a court without a source-anchored review pass first. Fluent and correct are not the same property, and treating them as interchangeable is how firms end up citing something that does not exist.

The pattern underneath all of this

Every use case above shares the same shape: Claude does the first pass on a task that is real work but not judgement work, and I review, correct and take responsibility for what comes out the other side. That is the whole doctrine of an AI-native practice, stated at the scale of one person's actual daily habits rather than a firm-wide policy. It is not exotic. It is just consistent, which is the part most people skip. I wrote the full sequence for building that consistency into a legal practice, not just an individual habit, in the AI-Native Practice Field Guide.