AI Receptionist
Set up 24/7 inbound call answering, lead capture, and booking requests.
The AI Receptionist answers inbound calls when you forward your business number to MaidPilot. It captures the caller's name, service need, address or property details, and preferred times, then creates a booking request for your approval. It does not auto-book your calendar.
Prerequisites
Before forwarding live traffic, complete these steps:
- Enable the AI Receptionist add-on or choose a receptionist plan.
- Set up your company profile, hours, and service area.
- Add the services and prices the AI should discuss.
- Configure knowledge, escalation rules, and greeting language under AI Secretary settings.
- Review the Dashboard numbers and call flow in a test call.
How to go live
Forward your existing business line to the MaidPilot number you are given. No new hardware or replacement phone number is required. Your customers can keep dialing the number they already know while MaidPilot answers the call.
Carrier forwarding options differ, so check the forwarding menu on your business line or speak with your provider. Keep your team aware of the change and publish the same number you always have.
What happens on a call
- The AI greets the caller and confirms their need.
- It asks for the details needed to create a booking request.
- It answers routine questions from your configured knowledge.
- It creates a request in the dashboard for you to approve or decline.
- The call is transcribed and summarized so you can follow up without replaying everything.
What happens after a call
Each call is transcribed and summarized in the dashboard so you can follow up on leads and approve or decline requests. Use the summary to update CRM notes, confirm the booking, or call back quickly when the AI handed off.
Sample call flow for a booking request
A successful call should feel simple to the customer. For example: AI greets, asks how it can help; caller asks about a standard clean for a 3-bedroom home; AI confirms date and time range, checks the service catalog, gives a starting price range, collects name and phone number, confirms the address and access notes, then says a human will confirm the booking. The result should be a request with enough detail to act on without another call.
Run a daily and weekly follow-up cadence
Check new requests and handoffs as part of the morning workflow. During the day, approve requests that fit capacity and reply to any caller who needs a callback. Once a week, review unanswered calls, missed opportunities, and calls where the AI gave an answer that did not match the current service list.
Troubleshoot forwarding quickly
If a test call does not reach the AI, confirm forwarding is still active, check that the forwarding target matches the MaidPilot number, and place a second test after waiting a minute. If only some calls reach the AI, look for carrier-specific forwarding settings, line busy rules, or a second phone line that customers may still be calling.
Design a useful booking request
Every booking request should contain enough information to confirm without another call: service type, property, preferred date and time, caller name, phone number, and any access note the caller volunteers. Review the first few requests and look for missing fields. If most callers forget an address or service detail, change the AI's question order before more traffic arrives.
Handle the first live week
Plan to review every call for the first week, even when the volume is small. Mark calls where the AI answered correctly, where the request needed a follow-up, and where the caller asked for something outside the knowledge document. Use those notes to improve wording, add a service detail, or create an escalation rule before scaling up.
Recover from a bad live call
When a call produces a wrong answer, do not treat it as only a one-off. Capture the exact caller question, the AI response, and the expected answer. Update the knowledge document or escalation rule, run a test with the same question, and confirm the improvement before the next call arrives. This routine keeps quality improving instead of waiting for a weekly review.
Define the minimum intake for a useful lead
List the fields needed to return a call and move the conversation forward: caller name, phone number, service type, property details, and a preferred date or time. If a caller will not share a detail, decide whether the AI should still create the lead or ask one polite confirmation question. The target is not a longer call; it is a request that can be handled without a second information-gathering call.
Review missed calls as a source of truth
When a call is missed or disconnected, note whether the problem was forwarding, a long silence, a question outside knowledge, or a caller who decided to wait. Group the notes by cause and fix the largest pattern first. Missed calls are useful feedback because they show where the answer or flow failed before a booking existed.
Set a weekly quality target
Pick a realistic target such as every non-escalation call ending with a clear next step, or every booking request containing the minimum intake fields. Review the past week against that target and update the question order or knowledge when the same field is missing repeatedly.
Schedule a quiet launch window
Forward live calls during a period you can review, such as a weekday afternoon or a low-volume day, rather than the start of a busy week. Keep one experienced person available to watch the first calls and fix an obvious greeting or knowledge error immediately. After the quiet window, expand to normal call volume only when the first requests look usable.
Worked example: an after-hours booking request
A customer calls at 7 p.m. and asks whether a cleaner can visit on Saturday for a move-out clean. The AI should greet, confirm the service, explain that a human will confirm the time, collect the address and move-out date, and create a request in the dashboard. On the next business day, the office opens the request, checks the Saturday crew and route, and sends a confirmation with the arrival window. The call does not promise exact arrival time or price beyond the approved starting guidance.
Define the operational metrics before launch
Pick simple numbers the team will review after go-live: calls answered, booking requests created, calls that needed a handoff, and callbacks completed within the target time. Record the baseline from the first quiet week, then review the next four weeks against it. If the team cannot name the number of calls that produced a usable request, the flow is not yet observable enough to tune.
Build a forward and answer checklist
- Confirm the business number still forwards to the MaidPilot target.
- Place one test call from the same number customers use.
- Verify a booking request reaches the dashboard with the minimum intake fields.
- Confirm the notification for a handoff reaches the right owner.
- Review the first three live calls and compare the AI wording with the policy.
- Update the knowledge document after every price, hours, or service change.
Review calls by outcome, not only volume
After the first month, group calls by the outcome: a useful booking request, a handoff that required a callback, a repeat question the AI still could not answer, or a missed call. Fix the largest group first. Volume alone can look healthy while the team is still calling everyone back to recreate the same information.