The intake call is the moment your firm knows the most about a prospective client's story and records it worst. The coordinator is listening, typing, and trying to keep the caller calm. The result is a page of fragmentary notes that the reviewing attorney has to decode before deciding whether to take the case. Details are missed, follow-up questions go unasked, and the notes that do survive are formatted differently by every person who takes calls.
Using an AI assistant to turn those notes into a structured summary fixes most of that in one step, provided the summary is designed to surface questions rather than answer them. This page shows the structure we use, the prompt that produces it, and the lines the tool must not cross.
From raw notes to a structured summary: the fields that matter
Design the summary before touching the tool. For most practice areas, a useful intake summary has the same skeleton: parties and contact details, a short narrative in chronological order, the key dates (incident, treatment, filing deadlines the attorney will need to calculate), the prospective client's stated goal, what documents the caller has, the referral source, and a list of open questions. The last item is the most important and the one hand-written notes never include.
Practice-area specifics go on top. A family law intake adds children, prior orders, and the other party's counsel if known, which also matters for AI search visibility for family law firms, since the questions clients ask an AI assistant before calling are the same ones your intake should be ready for. A personal injury intake adds insurance information and treating providers. An employment intake adds employer, dates of employment, and any agency filings. Write the skeleton for each practice area as a template, and keep it to one page.
A prompt structure that works for intake
Put the template and the rules into a Claude Project so the coordinator does not have to repeat them. The instructions say, in substance: you are converting intake call notes into a summary for a reviewing attorney; use exactly these headings; write the narrative in chronological order in plain language; put every date you find in the key dates section and mark any that were unclear; list every fact the attorney would need that the notes do not contain under Open Questions; do not offer legal conclusions, do not assess the merits, and do not calculate deadlines.
The coordinator pastes the raw notes and, if the call was recorded and transcribed with consent, the transcript. The summary comes back in the firm's format. The coordinator reads it against the notes, fixes anything wrong, asks the caller the open questions on the follow-up call, and files the summary in the lead record. That review step is not optional, and our page on AI hallucinations in legal work explains why a summary that looks complete can still contain a shifted date or an inferred fact.
Conflict checks, statute dates, and other things AI should flag but not decide
The summary can and should flag. It can list every party name so the conflict check is easy to run. It can note that an incident date is approaching a common limitations period and that the attorney should calculate the deadline. It can note that the caller mentioned a prior attorney, a bankruptcy, or an existing court order.
It should not decide. The conflict check runs in your case management system against your actual contacts. The limitations deadline is calculated by a person who knows the jurisdiction and the tolling rules. The merits are assessed by the attorney. Write these boundaries into the Project instructions and into the firm's AI policy so that a coordinator never treats a flag as a conclusion.
Connecting summaries to Clio Grow, Lead Docket, or Filevine
The summary should land in the lead record, not in an email. In Clio Grow, that means the lead's notes or a custom field, so it travels with the contact and matter when the lead converts to Clio Manage. Our Clio Grow intake guide covers the field setup. In Lead Docket, the summary sits on the lead and carries over to the Filevine project on conversion. In Filevine without Lead Docket, it goes into the Intake section of the project.
Once the summary has a fixed structure, the key fields (parties, dates, referral source, documents in hand) can be mapped to custom fields rather than left as prose. That is the step that turns a better summary into client intake automation: the fields drive the conflict check, the engagement letter merge, and the first task list.
What to do before the first live call
Test the Project on ten closed-file intake notes, redacted, and compare the summaries to what the attorney actually needed to know. Adjust the headings and the instructions until the open questions list is catching what was missed at the time. Then train every coordinator on the same Project, with the review step demonstrated, not described. This is the first workflow we set up in most Claude training for lawyers engagements, because it teaches the review habit every other AI use depends on.
Questions we get
Can the AI listen to the intake call directly?
Call recording and transcription tools exist, and a transcript is a better input than typed notes. Recording requires consent under the laws of the relevant states, and the transcript is client information that must go only into a tool your policy approves.
Will attorneys trust a summary the AI wrote?
They trust a summary the coordinator reviewed and signed off on, in a consistent format, with the open questions listed. The AI is producing the first draft; the coordinator owns the result.
How is this different from the intake features inside our case management software?
Some platforms now include AI features that summarize notes or transcripts. If your platform's feature produces the structure you need, use it and apply the same review step. The Project approach is worth building when you want control over the headings, the rules, and the open-questions logic, or when the platform's feature does not exist yet for your workflow.
If you would like a summary template and Project instructions written for your practice areas, tell us what you are working with.