Prospective clients contact law firms at night, at weekends and during their own working hours. An AI website assistant answers them immediately, qualifies the enquiry and captures intake so nobody waits until Monday. Done badly, the same assistant can give legal advice it should not or promise an outcome. Here is how to get the benefit without the risk.
What a good intake assistant does
- Answers general questions about the firm, its practice areas and how a consultation works.
- Asks structured intake questions: the type of matter, the other parties, how urgent it is and how to reach the person.
- Hands the lead to a human and saves it to the practice panel, so it is ready for a conflict check.
The guardrails
- No legal advice and no promised outcomes. The assistant explains the process and collects facts; it never evaluates the merits of a case.
- Matter-scoped context. It only knows what the firm has chosen to share, and nothing about other clients.
- Your own API key. The assistant runs on the firm's own Anthropic or OpenAI key. Keys are encrypted and never logged, and client data is never used to train models.
- Human in the loop. Any draft it prepares is marked as a draft for an attorney to review.
Confidentiality and your professional duties
Treat the assistant as you would a receptionist: it can take details and schedule, but it does not create an attorney-client relationship or give advice. Keep a short disclaimer on the chat window and review the captured intake before you act on it.
On the LegalWebDesigns platform the assistant is included with every site and follows these rules by design.