Family Law Client Intake: An AI Questionnaire and Urgency Triage

Use AI to collect the same core facts every time and to flag urgent signs, with a person making every decision. A good family law intake asks who is involved, what outcome is wanted, whether children are involved, any safety concern, and any deadline, then alerts a named person to urgent matters immediately.

The reason to build it carefully is the person on the other end. Someone filling in a family law enquiry is often frightened, angry or exhausted, and may be writing at midnight. An intake that feels like a tax form loses them, and an intake that misses a safety concern or a hearing date can do real harm. The aim is a questionnaire that is gentle in tone, short at the start, strict about escalation and quiet about advice.

Follow me on Instagram@sagnikteaches

Throughout, I am describing a workflow, not legal advice. Your professional rules on confidentiality, conflicts and client communications govern what you may do, so run any design past whoever is responsible for those at your firm.

Connect on LinkedInSagnik Bhattacharya

Decide what the AI is allowed to do

Write the boundary down before choosing a tool. For family intake I would give the AI exactly four jobs and forbid everything else.

Subscribe on YouTube@sagnikteaches

1. Ask the intake questions in a fixed order, in plain and kind language.
2. Summarise the answers into a one-page note for a person.
3. Flag urgency against a short list you define, and alert someone.
4. Offer next steps that are administrative: book a consultation, send documents, explain fees at a headline level you have approved.

It must not give legal advice, predict outcomes, say whether a case is strong, tell anyone what to do about a safety situation beyond showing emergency information, or promise that the firm will take the case. Put the same limits in the tool's instructions and on the screen the user sees.

The question set, in two stages

Ask little first. A first stage of six questions routes the enquiry. A second stage, after the firm accepts a consultation, gathers detail.

Stage 1: routing questions

1. What would you like help with? (Offer plain choices: separation or divorce, arrangements for children, financial settlement, a court order or a notice you have received, something else.)
2. Is there anything happening right now that worries you about your safety or a child's? (Yes, no, prefer to speak to someone.)
3. Do you have a court date, a deadline or a document that has to be answered soon? If so, when?
4. Are children involved? How many, and roughly what ages?
5. What is the other person's full name? (This feeds the conflict check, so say why you ask.)
6. What is the best way and time to contact you safely?

Question 6 matters more than it looks. If someone is living with an abuser who monitors their phone or email, a call or message at the wrong time can be dangerous. Ask whether it is safe to leave a voicemail, whether to use a specific number, and whether the firm's name may appear on the message.

Stage 2: the fuller questionnaire

Once the firm has agreed to a consultation, send the detail form. Group it so people can stop and return.

• People: full names, dates of birth, addresses, how long the relationship lasted, date of separation if any.
• Children: names, ages, where they live now, current arrangements, schools, any special needs, any existing orders.
• Home and finances: who owns or rents the home, main assets, debts, income of each person at headline level, any business interests, pensions.
• Legal history: any current or previous proceedings, orders, agreements, mediation attempts, and any lawyer already acting.
• Documents available: a checklist of what they can upload, such as a marriage certificate, recent statements and any court papers.
• Goals: in their own words, what a good outcome looks like in three months and in a year.

That list is a starting point. Which documents you need, and the facts relevant to your area of practice, differ by where you work and by the type of matter, so adapt it with your own lawyers.

Urgency flags: what the AI watches for

Define flags in advance and give each a route and a time target. The AI matches the user's answers and free-text to the list. When in doubt it flags upward.

The times in the right column are examples. Set ones your firm can actually meet, because a flag that is routinely missed is worse than none. Also decide who is on call outside office hours, and what the AI says when nobody is.

Wording for distressed people

Tone is half the product. A few rules I would build into the instructions.

• Short sentences. One question at a time.
• Acknowledge before asking. "I am sorry you are dealing with this. I will ask a few questions so the right person can help."
• Never say "I understand how you feel". The user knows it is a tool. Say what you will do next.
• Avoid legal jargon. Say "the other parent" rather than "the respondent".
• Let them skip. "You can leave any question blank."
• Always give a person: "If you would rather talk to someone, call [number]."

Illustrative exchange. The user types: "He said he will take the kids and I will never see them." The AI's reply, built from your script: "That sounds frightening. Are you or the children in danger right now? If so, please call [emergency number]. If not, I will make sure a member of our team sees this today. What is the safest way for us to contact you?" It then flags child removal risk, alerts the named person, and writes the exact quote into the note. It does not say whether he can or cannot do that.

Twenty-two weekend enquiries at a six-lawyer firm

Illustration only. Say a six-lawyer family practice receives 22 web enquiries over a weekend. Before AI, a receptionist read them on Monday, called the ones that looked serious, and sometimes found a hearing date mentioned in a fourth paragraph.

With the two-stage flow, each enquiry arrives as a short note. The AI summarises: type of matter, children yes or no, the other person's name, any date mentioned, the flag, and safe contact preference. Of the 22, the AI marks one as a safety concern, two with court dates inside three weeks, one with a financial emergency, and 18 as routine.

The duty lawyer sees the safety concern first and calls within the target time, using the number the person said was safe. The two court-date matters go to the lawyer who handles urgent applications. The receptionist books the 18 routine ones for the next fortnight. The conflict handoff, described next, runs alongside.

The time saved is mostly the receptionist's reading and rekeying, perhaps an hour or two on a Monday, and the real gain is that nothing urgent waits in a queue. The cost is review time: someone must still check the summaries and the flags, especially in the first weeks.

The conflict-check handoff

A conflict check cannot be an afterthought, and it must not be done by the AI alone. Treat it as a handoff.

1. The intake captures the full names of the enquirer, the other person, any new partner named, children, and any other party such as an employer or business.
2. The AI formats these into a list with spelling as given and alternative names or spellings, such as maiden names.
3. That list goes to the person who runs your conflict search against your client database. The AI does not search or decide.
4. Until the check is cleared, the AI tells the enquirer only that the firm will confirm it can act. It does not accept instructions, take sensitive details beyond the minimum, or say anything that could be read as an engagement.

That last point protects both sides. If you learn after a long questionnaire that you already act for the other person, you have collected confidential information from someone you cannot help. Keeping stage 1 short limits that exposure. Our tutorial on turning divorce intake notes into a chronology picks up once the matter is accepted.

Four more situations to script

The enquirer who is already represented

If they mention a current lawyer, the AI should record it and tell them the firm will be in touch about next steps, without discussing the matter. A person decides whether and how to proceed.

The enquirer who is the other party's friend or relative

Someone writing on behalf of another adult. The AI explains that the firm needs to hear from the person directly, and asks whether they would like a link to pass on.

The out-of-hours enquiry

At 2 a.m. the AI cannot offer a call. It confirms receipt, shows emergency information, states when someone will respond, and sends the urgent flags to the on-call person if you have one. Be honest about response times.

The enquiry about something you do not handle

A criminal charge, an immigration issue or a probate dispute. The AI says the firm does not handle that area and suggests the person seek a suitable lawyer, without trying to recommend a named firm unless your policy allows it.

A prompt skeleton for the instructions

Adapt this for the tool you choose. It is a skeleton, not a tested configuration.

You are the intake assistant for [firm]. You collect information for a lawyer to review. You are not a lawyer and you never give legal advice or predict outcomes.

Tone: calm, short sentences, one question at a time, no jargon. Never say you understand how someone feels.

Ask these questions in order: [stage 1 list]. Allow the user to skip any question.

Urgent triggers: if the user mentions violence, threats, fear for their safety or a child's, weapons or a child being taken, stop the questionnaire and show: "[emergency information]". Then record the exact wording and mark IMMEDIATE.
If the user mentions a court date, a served document or a deadline, record it and mark DEADLINE.

Never accept instructions, discuss the strength of a case or say the firm will act. Say a member of the team will review and contact them at [safe contact method].

At the end, produce a note: matter type, children, other person's full name, flags, quotes of any urgent statements, safe contact method.

Test it with at least twenty invented enquiries before real use, including awkward ones: angry users, minimal answers, people who write one long paragraph, and people who try to get legal advice out of the bot. Have a lawyer read the outputs.

A filled-in intake note, with what the lawyer does with it

The output that matters is the note a person reads in two minutes. Here is an illustrative one, invented for this tutorial, produced by the flow above.

INTAKE NOTE (AI-generated, for lawyer review)
Received: Sunday 22:40   Flag: DEADLINE + CHILD REMOVAL RISK
Matter type: arrangements for children (and separation)
Enquirer: [first name] [surname]   Other party: [full name] (also known as: [maiden/other names])
Children: 2 (ages 6 and 9), living with enquirer
Deadline: says a hearing is on [date], 17 days away; has a court document, not uploaded
Exact words about risk: "He said he will take the kids and I will never see them."
Immediate danger answer: No
Safe contact: text message only, after 09:00, do not use firm name
Conflict handoff: names listed above sent to conflicts clerk
Not asked: financial details (stage 2)

The lawyer reads the note, confirms the hearing date from the document when it arrives, and decides whether to call that morning. The "not asked" line is a small design choice that helps: it tells the reader which gaps are deliberate, so nobody assumes the AI forgot them. What the lawyer fixes: the AI wrote "separation" because the enquirer used the word once, but she had not said whether she was married, so the lawyer treats that as unknown.

Before and after: the same first message, two ways

A typical intake form says: "Please provide the full details of your matter, including the names and dates of birth of all parties, the nature of the dispute and any orders in place." Someone in distress reads that wall and closes the page.

The rewritten version asks: "What is the main thing you need help with? You can choose one." followed by the safety question, then the date question. The second takes about three minutes at the first sitting, and the rest waits. In a small firm, a before-and-after count of how many started enquiries were finished, week by week, is a good signal that the redesign works. Do not invent a target; measure your own baseline first.

Consent, privacy and where the data goes

Family intake holds some of the most sensitive information a firm sees: children's details, abuse allegations, finances. Settle these points before launch.

• Tell people what they are talking to. A line at the top: "This is an automated assistant. A member of our team will read your answers." Do not pretend it is a person.
• Say what happens to the answers. Who reads them, how long they are kept if the firm cannot act, and how to ask for deletion.
• Check the vendor. Where are the answers stored, does the vendor train models on them, which staff can access them, and can you delete a record on request? Get the answers in writing and file them.
• Collect the minimum at stage 1. You can always ask later. You cannot un-collect.
• No sensitive detail in notifications. The alert to your phone should say "urgent intake, see the system", not repeat an allegation in a message preview that others could see.
• Access control. Only people who need to read intake notes can open them, and conflict-sensitive details stay out of general channels.

Training the people who receive the flags

The software is the smaller part. The people who answer a safety flag need a short script and practice. A 30-minute session each quarter is enough to start.

1. Role-play a call-back to someone who can only talk for two minutes.
2. Agree what staff say when asked for advice before a conflict check clears.
3. Agree who may offer emergency information and who escalates to a lawyer.
4. Practise using the safe contact method as recorded, including what to do if the call goes to someone else.
5. Review one anonymised flagged note together and ask what you would have done differently.

Write the outcome into a one-page procedure and keep it where the duty person can find it at 9 a.m. on a Monday.

Time and cost: a rough sum for a small firm

Illustration, with invented figures to show the method. Say a firm gets 90 enquiries a month and a staff member spends 12 minutes reading, copying and chasing each one: 18 hours a month. If the flow cuts that to 5 minutes of review per enquiry, it saves about 10 hours a month. Multiply by your internal hourly cost and compare with the tool's monthly fee and the set-up time for scripts, testing and training. Include the review time in the sum, and include the cost of a missed urgent matter as the reason you keep the humans in the loop. If the saving is small, the real value may still be the consistent flagging, but be honest about which one you are buying.

How to check it works after launch

• Review every flag for the first month. Read all the notes where the AI marked something urgent, and a sample of those marked routine. Look for missed urgency, because a false alarm costs a phone call and a miss costs far more.
• Time the response. Measure from enquiry to a person's first contact for each flag type, and compare with your target.
• Read the drop-offs. If many people abandon at one question, rewrite it or move it to stage 2.
• Collect feedback. Ask new clients how the first contact felt, in one line.
• Audit the data. Confirm answers are stored where they should be and deleted on schedule when a matter is declined.

If you would like to design this with someone, that is the sort of job a 1:1 call covers, and you can reach me through the contact page. For the phone side, read about AI receptionists for family law, and once a client signs on, our tutorial on drafting a family law retainer with AI covers the next paper.

Questions firms ask before putting AI in front of family clients

Should AI ever decide whether a matter is urgent?

It should flag, not decide. Let the AI spot trigger words such as threats, a child at risk, a hearing date or a missed deadline, and route those to a person within a set time. A lawyer or trained staff member decides what the urgency means and what happens next.

What should the AI say if someone describes immediate danger?

It should stop the questionnaire, show emergency contact information, and alert a named person at the firm. Do not let it give legal advice or keep gathering details. Write this script yourself, with your own region's emergency numbers, and test it before launch.

Can the intake bot tell people what their rights are?

No. Keep it to collecting facts, explaining the next step and what the firm does. Anything that reads like advice creates risk, and a distressed person may rely on it. Tell users plainly that a lawyer will review their answers.

How do I stop the questionnaire becoming too long?

Split it. Ask only five or six items to route the enquiry, then send a fuller questionnaire after the firm has accepted a consultation. Long forms at the first contact lose people in distress, and they also collect more sensitive data than you need before you know you can act.

Where should the answers be stored?

In your practice-management or case system under your confidentiality and retention rules, not in a loose chat history. Check where the intake tool stores data, whether it trains on it, and who at the vendor can see it, before any real enquiry goes through.

Further reads

• Best Legal AI for Family Law: What to Pay For and What to Skip — What to pay for and what to skip in legal AI.
• AI for Family Law Attorneys: Summarise Financial Disclosure Safely — Summarising financial disclosure safely after intake.
• AI Voice Agent for Lawyers: Cost and Safe Handling of Intake Calls — Voice agents for intake calls, with cost and limits.
• Family Law Client Communication With AI: Calm, Consistent Emails — Calm, consistent emails after the intake stage.
• AI Legal Chronologies for Family Cases: Events Linked to Evidence — Linking events to evidence once the matter opens.
• Personal Injury Lawyer Chatbot: Intake Questions, Limits and Cost — A comparison from another practice area.
• Best AI for Family Law in Australia: Privacy, Court Rules and Cost — A family law shortlist with prices, what the Family Court's AI practice direction requires, and the privacy rules that keep client data out of free tiers.
• Conflict Check Software for Small Law Firms: Where AI Helps — Where AI helps in a conflict check, where it must not decide, and what a partner signs off, with a worked example and a 20-minute test of your own tool.
• Family Law Document Automation: Gavel Workflows or Generative AI? — Rules-based assembly against prompting for family law documents: costs for a four-fee-earner firm, how each fails, and a hybrid that uses both.
• Lawmatics Merlin: Is It Worth Adding Qualify, Engage and Copilot? — What Merlin Qualify, Engage and Copilot do for law-firm intake, what each costs, and a four-week test on real leads before you pay.
• Legal Client Intake Software: What to Compare Before You Choose — How a small law firm should compare intake software with AI: the AI's actual job, the conflict-check step, integrations and a 20-enquiry test.
• Best AI for Family Law UK: Form E Prep and Client Updates — A safe route to Form E bank statement summaries and calmer client updates: what to redact, which tools to use, what to check and what the SRA has said.
• AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.

Sources: general intake and triage practice for family law firms; no vendor pricing or features relied on. Confirm confidentiality and conflict rules with your regulator or professional body.