Skip to content
be-human.ai
← Back to the journal

AI by industry

AI for Family Law Firms: Intake Screening and Court Dates

How a family law office can let an AI agent handle intake forms and court-date reminders while attorneys keep conflict checks and legal advice.

Be Human typographic cover with the title: AI for Family Law Firms: Intake Screening and Court Dates
Conceptual illustration created with AI · Be Human

An AI agent can run the administrative side of a family law intake desk safely: it asks the firm's approved questions, collects documents, tracks what is missing, and sends court-date reminders from a calendar that staff verify. Conflict checks, legal assessments, and any statement about what a prospective client can expect stay with licensed attorneys, and the firm writes that boundary down before the agent goes live. This article explains where the line sits and how to set it up.

Key takeaways

  • An intake agent collects and organizes facts. It does not assess the merits, advise on options, or decide whether the firm will take a matter.
  • Conflict checks stay with attorneys or a designated conflicts person. The agent only formats names and routes them into that process.
  • Court-date reminders should read from the firm's docketing system or the attorney's record, with the agent acting as messenger and tracker.
  • Memory, document access, and permissions are scoped to the role, so the agent sees what its job requires and nothing more.
  • A useful first engagement names one task, one approval boundary, and one observable success criterion that the office can check each week.

What an intake agent can and cannot decide

Family law intake gathers the facts that determine whether the firm can act, and those facts arrive in messy form: a message late at night, a half-finished form, a name spelled two ways in two places. An agent can make that stage faster and more complete. It can ask the questions on the firm's approved form, check that required fields are present, file documents into the right matter folder, and tell the prospective client what happens next.

It should not interpret custody standards, estimate financial outcomes, or tell someone whether they have a case. Those are legal judgments, and they belong to people who carry professional responsibility for them. Put the line in three places: the written intake policy, the agent's instructions, and the permissions the agent actually has in the systems it uses.

A step-by-step plan for intake forms

Start with the form, not the software. The supervising attorney should approve every question, mark which answers are required before a consultation can be scheduled, and decide which answers may be left as declined with a recorded reason.

  • Write the questions in plain language and lock the approved wording. The agent should not paraphrase legal terms or offer examples of how a court has ruled.
  • Define stop conditions before launch. Signs of immediate danger, a request for emergency relief, a deadline within days, or a person in acute distress should route at once to a named staff member, following the firm's own safety protocol.
  • Set a completion rule. A file moves to attorney review only when required fields are complete or marked declined, and the reason is recorded.
  • Decide what the prospective client is told at the end, in plain language. State that submitting information does not create an attorney-client relationship, and have an attorney approve that sentence.
  • Review a sample of finished files each week against the approved form to confirm the agent is following the script.

The stop conditions matter most. The agent should never be the party that decides a situation is low urgency. Its job is to notice and hand off. A similar intake limit appears in clinical settings, as described in AI for Behavioral Health Clinics: Intake and Scheduling Limits, and the same discipline applies here.

Conflict checks: the step that stays with attorneys

A conflict check compares the names in a new matter, including the opposing party, children, employers, and related parties, against the firm's records of current and former clients. The duty belongs to the firm, and so does the judgment. The agent can help by collecting a complete list of names in a consistent format, including alternate spellings the prospective client supplies, and by placing that list in the conflicts queue.

It should not run the check and report it clear, and it should not tell anyone that the firm has no conflict. Clearing is a decision a person makes and records. A workable pattern looks like this:

  • The agent extracts names from the completed form into a structured list, labeling each role such as opposing party or child.
  • The list goes to the conflicts queue, not back to the prospective client as a confirmation.
  • The prospective client receives a neutral message that the firm is reviewing the request, with no timeline the firm has not approved.
  • Only the designated person can set the status to cleared or declined, and that action carries their name and the time.
  • The agent cannot offer a consultation slot until that status is set by the designated person.

Each jurisdiction has its own ethics rules on conflicts, screening, and client communication. Confirm the firm's obligations with its bar or ethics resource before building the routing.

Court-date reminders that do not become the calendar of record

Reminders are among the lowest-risk tasks for an agent, provided it is a messenger rather than a source of truth. The authoritative date lives in the firm's docketing system or in the attorney's record, and staff verify it against the court's notice. The agent reads from that source, sends reminders at set intervals to the right person, records confirmations, and reports changes.

It should not infer a date from a message that says the hearing is next month, because the court notice controls. Reminder wording also needs review: it should contain no strategy, and it should direct the client to the official notice or to the attorney's message for details.

Illustrative scenario (not a real matter): a paralegal enters a hearing into the docketing system after a notice arrives. The agent sees the entry, sends the client a reminder a week ahead with the time, location, and a request to confirm, and flags the matter to the paralegal if no confirmation arrives in time for follow-up. The attorney receives that flag, not the agent's guess about whether the client will attend.

Transcript orders and similar post-hearing records follow their own workflow, covered in Superintelligence Prep for Court Reporting Firms: Transcript Orders.

Memory, documents, and access limits

Family matters contain some of the most sensitive records a firm handles: financial statements, medical notes, messages about children, and allegations that may involve harm. The agent's memory should hold only what its role needs, such as intake status, the next known date, and the list of open questions for the attorney. Full document contents should stay in the firm's document system unless the firm has decided, in writing, that the agent may hold them.

  • Keep: intake status, required-field gaps, the next verified date, reminder history, and the handoff log.
  • Do not keep: full financial records, medical details beyond what the form requires, or the text of sensitive messages from the other party.
  • Grant access per role through each system's own permission model, and make the agent's credentials easy to revoke.
  • Treat everything a prospective client types as untrusted input, since anyone can paste instructions into a free-text field. The OWASP guidance on prompt injection describes this risk and the defenses that address it.
  • Review permissions on a fixed schedule and after any change in staff or systems.

For the risk review itself, the NIST AI Risk Management Framework gives a practical structure: map each use and its failure modes, measure how the agent behaves against them, and assign an owner to every control. Data protection duties depend on where the firm and its clients are, so confirm them locally, including the rules that apply to Portuguese-speaking clients if the firm serves them.

Starting small: the first engagement and its success criterion

The first step is rarely the whole intake system. It is one task in one practice area, with clear edges.

  • Which single task will the agent own first, such as completing intake forms or sending reminders for confirmed hearings?
  • Who approves each boundary, and who receives each handoff?
  • What observable result will the office check each week, such as the share of files that reach attorney review complete, or the number of reminders that went out from a verified date?
  • What does the agent do when it cannot finish, and where does that case go?

Be Human designs and operates this kind of setup as managed infrastructure. A first engagement defines one role, the memory and tools that role needs, the approval boundaries the attorneys set, and an observable success criterion. Ongoing operations can include monitoring, recurring evaluations of the agent's answers against the approved forms, permission reviews, incident response, and controlled changes when the firm's procedures change. Service is remote, in English and Portuguese.

If you want to find the first task in your office worth handing to an agent, take the free Be Human AI diagnostic.

Frequently asked questions

Can an AI agent run intake for a family law firm? It can handle the administrative parts: asking approved questions, collecting documents, tracking missing fields, scheduling, and sending status updates. Legal assessment, conflict clearance, and any statement about likely outcomes stay with licensed attorneys. The firm defines that boundary in writing, and the agent operates only inside it.

Who is responsible for a conflict check that involves an agent? The firm is responsible. The agent gathers names in a consistent format and routes them to a designated person, who makes the decision and records it. The firm should also confirm its duties under its jurisdiction's ethics rules before the routing goes live.

Should court-date reminders come from the agent's own calendar? No. The authoritative date should come from the firm's docketing system or the attorney's record, verified by staff against the court's notice. The agent reads that source, sends reminders, tracks confirmations, and flags any missing confirmation to a person.

How do we start without launching a long project? Choose one task in one practice area, write its approval boundary and stop conditions, and define one success criterion the office can check each week. Keep the agent's memory and permissions limited to that task, then expand only after reviews show it behaves as the policy requires.

Your next step

See where an AI agent fits in your company first.

Take the free Be Human AI diagnostic. A few questions about your work show which task to hand to an agent first, what it needs to know and where a person approves.

Be Human editorial guidance. Examples are illustrative unless explicitly attributed. Sources provide background; implementation choices depend on your workflow.

Less busywork. More room for you.

Bring us the work you wish you had time for.

Book a call to try a personalized AI agent free for 7 days ↗Or take the free AI diagnostic →