AI Intake Chatbot for Attorneys

Where prospective clients slip away before you ever speak

Ask an attorney where new matters come from and you hear about referrals and marketing. Ask where they are lost and the answer is almost always intake: the gap between someone reaching out and someone at the practice actually engaging with them. People with a legal problem commonly contact more than one lawyer, and the practice that responds usefully first has a real advantage.

It is not a replacement for answering the phone. Callers still reach your staff or answering service; the bot covers the website, and optionally text messages, where the leaks above are worst.

Turning your case-acceptance criteria into rules the bot follows

Every attorney has acceptance criteria, even if they have never written them down: which states, which kinds of matter, how recent, whether the person already has a lawyer, which opponents you won’t take on. The core of an intake build is writing those criteria down with you and turning them into rules with a small set of outcomes:

• Book: the matter plainly fits, so the bot offers consultation times.
• Attorney review: it might fit, so the intake goes to you before anything is promised.
• Refer out: it clearly does not fit, so the bot explains that and offers your chosen referral path.
• Call now: something time-critical, handled in its own section below.

An illustration of how that looks for a hypothetical solo practitioner who takes motor-vehicle injury cases in one state:

An estate planning practice looks completely different: in-state residents who want a will or trust go straight to booking, a probate question after a death goes to review because court timing varies, and a contested estate is referred out if the practice does not litigate. A criminal defense practice might book anyone charged in the counties it covers and send federal matters to review. These are examples of structure, not advice about which matters to accept.

The prospect never sees the rule or a verdict about their situation, only the next step. The cutoffs, the covered counties and the exceptions are yours; we never supply a limitation period or deadline from our own assumptions.

The intake record: each field, and why it is asked

A good intake asks nothing without a reason, and asks it at the right moment. This is the default record we start from, adjusted for each practice:

The opposing-party row sits above the description row on purpose. Duties to prospective clients under ABA Model Rule 1.18 and the state rules modeled on it mean the practice is better off learning who is on the other side before learning much else. How those duties apply to you is a question for your ethics counsel; our job is making sure the order holds in every conversation, including the ones that start with a long paragraph of facts.

What reaches the attorney before the consultation

The payoff for the attorney is walking into a consultation already knowing the basics. Each completed intake produces a one-screen summary attached to the lead record. A fictional example:

Three things are deliberate. Everything is labeled as reported by the prospect and unverified. The summary records which rule fired, so when an outcome looks wrong you can see why and correct the rule. And it contains no legal analysis: no guess about liability or value, and no list of possible claims. That thinking belongs to the attorney, and a summary that pre-judges it invites anchoring.

Feeding your lead pipeline in Clio Grow, Lawmatics or Lead Docket

Intake software organizes prospects into pipeline stages. The bot’s outcomes map onto those stages so nothing lives only in a chat log:

On the intake side, the tools we see most are Clio Grow, Lawmatics and Lead Docket. On the practice-management side, Clio Manage, MyCase, PracticePanther and Smokeball; personal injury practices often run Filevine, Litify or CASEpeer. Consultations go onto Outlook, Google Calendar or Calendly, and paid consultations use the LawPay link your practice already has. Engagement agreements stay with your staff, typically through DocuSign, after the attorney decides to take the matter.

Some specifics we handle in the integration: duplicate detection when the same person writes twice, a source field so you can see which marketing produces accepted matters, and a strict one-way flow in which the bot creates leads but never searches existing client files. Where a product lacks a usable API, we say so before quoting and propose a fallback, such as a structured email the software can import.

Intakes that stop halfway

People leave mid-intake all the time. A phone battery dies, the names question makes someone nervous, or they need to find a date in their paperwork. Because contact details come first, a partial intake still becomes a record, marked incomplete and showing exactly where the person stopped.

What happens next is your policy, not ours. Following up with someone who contacted you is different from approaching a stranger, but rules on lawyer communications vary by state, and texting in particular requires consent the practice records properly. We build whatever you decide, whether that is one call from staff, one message, or nothing at all, and we don’t build automated nag sequences.

We also avoid a “resume where you left off” link by default. On a shared family computer or phone, that link can expose what someone typed to whoever picks up the device next. If you want resumable intakes, the prospect confirms a code sent to their own phone or email first.

The call-now outcome, and its limits

Call-now triggers are set by you: a person in custody, a hearing within the number of days you choose, a deadline you consider imminent. For a solo practitioner the usual design is a text alert to the attorney’s phone with a two-line summary and a callback number; for a larger practice, an on-call rotation. If someone describes being in danger, the bot shows emergency guidance, starting with 911, before anything related to the practice.

Recognition is imperfect. We test trigger detection with many phrasings, misspellings and Spanish, yet a bot can still miss an urgent message worded in an unexpected way. So staff review every intake that mentions a court, a jail, a deadline or safety during business hours, flagged or not, and the bot never promises a callback time your practice can’t honor.

Declining well: referrals and the record you keep

Many inquiries fall outside what a given practice does. The bot says so directly and without any comment on the merits, for example: “This practice doesn’t handle bankruptcy, so we can’t help with this matter.” It then offers the path you choose, usually the state or local bar association’s lawyer referral service, or legal aid for someone who mentions they can’t afford a lawyer. Where you want it, it adds a neutral reminder that legal deadlines may apply and speaking to a lawyer promptly is sensible.

Two decisions belong to your attorneys. First, whether a declined chat needs a formal non-engagement letter; if so, the bot logs the decline for staff to send one. Second, how long declined intakes are kept. Many practices keep at least the party names so future conflict checks can see the prior contact, and we build retention to match what you decide.

What the bot keeps, where, and who reads it

• Minimize first. Length caps on free text and the on-screen instruction not to share confidential detail reduce what is collected in the first place.
• One system of record. The intake lives in your intake or practice-management software; any copy elsewhere is short-lived and has a retention period you set.
• Model provider settings. We use language-model services under business terms that keep your conversations out of model training, and we give you those terms to review. ABA Formal Opinion 512 on generative AI is a useful reference for that review.
• Named access. Transcripts are visible only to roles you designate, each person with an individual login, and operational logs record events rather than what prospects typed.

We make no compliance or certification claims, and nothing here states that a configuration meets your jurisdiction’s rules. The practice supervises the bot as it would any assistant, and we give you the flow, wording, data map and vendor terms in writing so that judgment can be made properly.

Numbers that show whether intake is working

Chat volume says little about a legal practice. The measures we set up with you instead:

• Consultations the attorney judged a genuine fit, rather than consultations booked.
• Drop-off by question. A spike at the opposing-party step usually means the reason for asking needs a clearer explanation on screen.
• Time from completed intake to cleared conflict check.
• Rule outcomes the attorney overrode, which show where the acceptance rules need refinement.
• Declines that left with a referral, so turned-away people still had somewhere to go.
• Accepted matters by source, which tells you where marketing effort pays.

What we need from your practice, and what you get back

The first step is a free phone or video conversation, and scope is fixed in a written quote before any build work begins. The people doing this work sit in the Houston area; EVOTECH IT LLC dates back more than 20 years, carries a 5.0 average on Google, and can serve attorneys in any state remotely. Practices that also want chat for existing clients and general website visitors should read our page on AI chatbots for law firms. To walk through your current intake, phone (832) 359-2425.

Related services

Frequently asked questions