At a glance
- A fixed field map sets what every intake captures, where it comes from, and who confirms it before it becomes a matter record.
- An intake agent follows a scripted boundary: a short list of things it may say, and a longer list it must never say to a prospective client.
- Routing by practice area and urgency gets the inquiry to the right attorney queue; an "other" or ambiguous inquiry goes to a human, not a guess.
- Case acceptance and any assessment of the legal issue stay with attorneys; the system's job ends at a complete, routed record.
- Declined matters get a reason, a date, and a notification on record — they don't just disappear from the queue.
What should an AI intake capture before an attorney calls back?
AI legal intake automation answers one question before a human picks up the phone: what does the attorney need for a fast, informed first decision? The system's only job is capture — turning the caller's own words and contact details into a record the attorney can act on.
- Who is asking, and every reliable way to reach them.
- Every other party named in the story, including companies and affiliates, so a conflicts check has something to search.
- The practice area and a short, specific description of the issue, in the prospective client's own words where possible.
- Any date mentioned, captured verbatim rather than interpreted, so the attorney sees it first.
- How the inquiry arrived and how they found the firm.
- Consent to the firm's intake terms and privacy notice, with a timestamp.
Case acceptance and any assessment of the issue stay with the attorney; nothing in the record is a decision.
The inquiry-to-matter-record field map
Adapt the required column to your own intake policy. Each row needs a source, a validation rule a form or agent can enforce, and a named role who confirms it.
| Field | Source | Required? | Validation | Who confirms |
|---|---|---|---|---|
| Prospective client name | Form, chat or call transcript | Required | Matches a reachable contact method; no blank submissions | Intake staff |
| Contact details (phone, email, preferred method) | Form or chat | Required | At least one valid, reachable channel | Intake staff |
| Opposing party / other-side names | Form or chat | Required | Full legal name requested, not just a nickname or "the other guy" | Conflicts review |
| Related entities or affiliates mentioned | Form or chat, free text | Optional, flagged if present | Passed to the conflicts query as written, not summarized | Conflicts review |
| Matter type / practice area | Form picklist, or classified from free text | Required | Matched to the firm's practice-area list; anything that doesn't match routes to a human | Intake director |
| Short description of the issue | Free text or chat transcript | Required | Minimum length; too short is flagged for follow-up, not guessed at | Attorney |
| Jurisdiction or venue | Form picklist | Required where known | "Unknown" is a valid value, not a blocked field | Attorney |
| Key dates mentioned (incident, notice, deadline) | Extracted from free text or chat | Required if mentioned | Captured verbatim, never interpreted as "too late" or "in time" | Attorney |
| Referral source | Form picklist plus free text | Required | At least one value recorded | Intake staff |
| Fee arrangement mentioned by the caller | Chat or call notes | Optional | Logged as the caller's stated preference, never as a quote from the firm | Attorney |
| Prior contact with the firm | Automated lookup against existing client/matter records | Required | Matched by name and contact details, not name alone | Conflicts review |
| Documents offered or uploaded | Intake portal or file request | Optional | Logged by file name and date; type-checked, not opened or read by the agent | Intake staff |
| Urgency flag | Derived from captured dates against a firm-set window | Required | Set automatically; never left to the caller to self-report | Attorney sees flagged items first |
| Attorney decision (accept/decline) with timestamp | Attorney review of the completed record | Required | Timestamp set automatically when the decision is recorded, not estimated | Attorney |
| Decline reason, date and notification date | Attorney decision record and the notification log | Required if declined | Notification date logged only once the prospective client is actually notified, not when drafted | Intake staff |
Intake and practice-management tools can already handle parts of this. Clio publishes API reference documents for its Clio Manage and Clio Grow products [1]. Lawmatics lets a firm define a custom intake form and the fields to collect by practice area and case type [2], and tracks each lead through the stages of the client journey in one place [3].
What must an intake agent never say to a prospective client?
An intake agent collects information; it doesn't evaluate a case, quote a fee, or confirm who the firm represents. The table below is a draft boundary: what the agent may say to move the conversation forward, and what it must never say because it would pre-empt an attorney's decision.
| Topic | The agent may say | The agent must never say |
|---|---|---|
| Purpose of the call | I'm collecting information so an attorney can review your situation. | I'm your lawyer / We represent you (before any engagement exists). |
| Case merits | I can't evaluate your situation — that's for the attorney to assess. | You have a strong case / You'll probably win / You don't have a case. |
| Fees | I'll note what you're looking for, and the attorney will discuss fees with you. | Any specific fee, discount, promise about total cost, or term of a fee agreement. |
| Deadlines | I'll record any date you mention so the attorney sees it first. | Whether a deadline has passed, or what that would mean for the matter. |
| Existing-client or conflicts status | I'll run your information through our standard review before we go further. | Confirming or denying that another named party is a current or former client. |
| General legal questions | Clarifying intake questions only, in the caller's own words. | Any answer to "should I…", "do I need…" or "what are my rights…" questions. |
| Confidentiality of what's shared | What you share goes into our file for the attorney's review. | A specific promise about privilege or confidentiality protection before engagement. |
| Timing of a response | Here's our stated window for a first response, offered as information. | A promised callback time or a commitment that a specific attorney will take the matter. |
How should inquiries be routed by practice area?
Routing is where legal intake triage happens: turning a record into a destination a team actually watches. The practice-area field drives the default; a few other signals override it.
| Practice area signal | Default destination | What overrides it |
|---|---|---|
| Personal injury | Personal injury intake queue | An existing matter for the same client routes to that matter's attorney instead |
| Family law | Family law intake queue | An existing-client match routes to the client's current attorney |
| Corporate / transactional | Corporate intake queue | An existing engagement for the named entity routes to that matter's attorney |
| Immigration | Immigration queue, split by case type if the firm tracks it | An existing-client match routes to that matter's attorney instead; otherwise the case-type field decides |
| Missing or ambiguous practice area | Human triage queue, not an automatic matter record | An intake director reassigns it to a practice group by hand |
The urgency flag from the field map rides along with routing: a flagged inquiry should reach the front of its queue, not just the right queue.
How do you measure intake speed and quality?
Measure the record, not the conversation. Each metric below is a method for your own intake data. This section gives methods, not benchmarks.
| Metric | How to compute it | Watch out for |
|---|---|---|
| Time-to-decision | Timestamp of first capture to timestamp of the attorney's accept/decline decision | Stopping the clock at "routed" instead of "decided" overstates speed |
| Intake completion rate | Completed records divided by started intakes, by channel | Form abandonment and dropped calls are different failure modes; track them separately |
| Data-completeness rate | Required fields present divided by required fields in the field map, per record | Decide whether "unknown" counts as present before you measure it that way |
| Callback-window adherence | Flagged inquiries first-touched inside the firm's stated window divided by total flagged inquiries | An urgency flag set too broadly makes this easy to hit and meaningless |
| Declined-matter turnaround | Decline decision timestamp to closure-and-notification timestamp | A decline that never gets a recorded outcome looks the same as a live file |
Where to start: a first scoped workflow
- Pick one queue
Choose a single practice area and intake channel — for example, the web form for one practice group — rather than every channel at once.
- Build the field map for it
Adapt the table above to that queue only: which fields, which validations, which role confirms each one.
- Draft and route the script boundary
Write the allowed/not-allowed statements for that queue, and send it to a reviewing attorney and your compliance contact for sign-off before it's used live.
- Automate capture and routing
Wire up the form or chat to populate the record and route it to the chosen queue; leave anything ambiguous to the human triage queue.
- Run it alongside the existing process
For a set number of inquiries, compare where the system routed each one against where a person would have sent it, and fix the mismatches.
- Add measurement and the declined-matter path
Once the queue is stable, turn on the metrics above and the decline-and-notify workflow, then expand to the next practice area.
When this doesn't apply
- Inquiry volume is low enough that a person already reviews and routes every one without delay — the field map is still useful on paper, even without automation.
- Nearly every inquiry in a practice area needs an immediate attorney judgment call; that gains little from automated routing beyond the field map and a human queue.
- Your conflicts process isn't ready to consume a structured intake record yet. Fix that sequencing first, since automation increases the volume hitting it.
- The firm hasn't agreed on a practice-area list or routing ownership. Settle that before building the routing rules, not while building them.
Frequently asked questions
What should an AI intake capture before an attorney calls back?
At minimum: who is asking and how to reach them, every other named party so a conflicts query has something to search, the practice area and a short description of the issue, any dates mentioned, how they found the firm, and their consent to the firm's intake terms. The attorney should inherit a complete record, not a transcript.
What must an intake agent never say to a prospective client?
It must never evaluate the merits of a case, quote or promise a fee, confirm or deny who the firm represents, answer a direct legal question, or promise confidentiality terms before engagement. A reviewed script boundary sets what it may say instead: that it's collecting information for an attorney to review.
How should inquiries be routed by practice area?
The practice-area field captured at intake drives a default queue for each practice group, with urgency flags moving a record to the front of its queue. An existing-client or existing-matter match overrides the default and routes to that matter's attorney; a missing or ambiguous practice area goes to a human triage queue instead of a guess.
How do you measure intake speed and quality?
Track time-to-decision from first capture to the attorney's accept or decline, intake completion rate by channel, the share of required fields present at hand-off, adherence to the firm's stated callback window for flagged inquiries, and how long a declined matter takes to close out and notify the prospect.
What happens to declined matters?
A declined inquiry is marked with a reason and a date, the prospective client is notified in writing under the firm's own process, and the record is kept under the firm's file-retention policy. Keeping the record lets a later, related inquiry be checked against it instead of starting from nothing.
Related reading
- The next step for this workflow is customer support escalation, which covers triage rules and handoffs with context once a request is already moving through a queue.
- For the document-handling side of a matter once it's open, see document review with human approval.
- Our article on AI document intelligence for legal and compliance covers the document-review workflow further downstream from intake.
- For the general principle behind keeping a person in the loop, see human-in-the-loop AI governance.
- For a broader view of AI across law firm operations, see our legal industry page.
Sources
- Clio, "Clio Developer Documentation Hub" (retrieved 2026-10-01) (accessed 2026-10-01)
- Lawmatics, "Client Intake Software for Law Firms" (retrieved 2026-10-01) (accessed 2026-10-01)
- Lawmatics, "Legal CRM Software for Law Firms" (retrieved 2026-10-01) (accessed 2026-10-01)
More in Professional firm operations
AI Proposal Automation for A&E Firms: Speeding Up SF330 and RFP Responses Without Losing Quality Control
AI proposal automation for A&E firms: a workflow table showing what an agent drafts for SF330 and RFP responses and what your licensed staff review.
Engineering FirmsWhy Clients Reject Your Invoices: Reducing Billing Disputes at Consulting and A&E Firms
Why clients reject your invoices, and how an AI review catches fee-cap overages, wrong phase codes and missing backup before billing. Includes a cause table.
Engineering FirmsWIP Accounting for A&E Firms: Catching Overbilling, Underbilling and Subconsultant Reconciliation Gaps
WIP accounting for A&E firms: what a WIP report shows, how to spot over- and underbilling, and a weekly exception table that includes subconsultant invoices.
Engineering Firms