AI answers the first questions so a person can answer the real ones

A new inquiry arrives with half the information a person needs and none of the context. The lead-response agent handles that gap: it replies while the lead is still reading your page, confirms what they asked for, and collects the details your team would otherwise chase by phone. Everything it says comes from answers you approved during installation.

• Replies to form, text, and chat inquiries on arrival
• Confirms the service asked for in the visitor's own words
• Collects job type, location or vehicle, and timing
• Books an approved time slot or hands the thread to a named person

The agent works from a short required-field list per service. If a detail is missing it asks for that one thing, records what it got, and still creates the record and the alert. It does not hold a lead hostage behind a questionnaire, and it does not guess a value into a field.

• One follow-up question per missing required field
• Partial records still route to a person with the gap flagged
• No invented addresses, vehicles, dates, or measurements

It replies on the channel the person used. Text replies require the number to have been given for contact, quiet hours are respected, and any opt-out wording stops messaging on that channel immediately and marks the record. Approved wording is written during installation and stored, not improvised per conversation.

• Same-channel reply: form to email or text, text to text, chat to chat
• Quiet hours honoured on outbound messages
• Opt-out stops the channel and flags the record
• Approved message library reviewed with your team before launch

Price questions, complaints, warranty issues, insurance work, anything urgent, and anything outside the approved topics escalate. The agent says a person will follow up, stops answering, and notifies the owner of that record.

• Price, negotiation, and scope questions escalate immediately
• Complaints and warranty language escalate and pause automation
• Repeated unrecognised questions escalate rather than repeat themselves
• Every escalation notifies the assigned person on the record

Automation stops on a human reply from your team, on a booked appointment, on an opt-out, on an escalation, or after the agreed number of attempts with no response. A stop is visible on the record, so nobody is ever guessing whether a sequence is still running.

When AI lead response is part of the agreed scope, Cross X Agency writes the required fields, approved answers, escalation triggers, and stop conditions with your team, then tests the paths before launch. It is never assumed to be active in every standard setup.