01 · Trigger
The customer asks from the AI they already use
The request arrives with the service type, address, urgency, and preferred window. No new app or account is required for the customer.
Workflow demo · home services
Turn an after-hours service request into a dispatch-ready job
The lead should reach Jobber before it reaches voicemail.
A concrete Business MCP workflow for HVAC, plumbing, electrical, cleaning, and other field-service shops. The AI collects the facts, checks the real schedule, and creates the handoff without inventing a price or promising a technician.
ChatGPT, Claude, or GrokJobber / Housecall ProCalendarOwner approval queue
A homeowner asks ChatGPT for help with a leaking water heater at 8:17 PM. The shop is closed, but the booking intent is live.
The AI moves the request. Your rules decide how far.
02 · AI action
The connection asks only the approved intake questions
It separates an active leak or safety issue from a routine estimate, confirms the service area, and checks live availability.
03 · Existing system
A clean request lands in the field-service system
The contact, address, issue, source, and requested time are written to Jobber or Housecall Pro. The system of record stays the system of record.
04 · Human boundary
Price promises and exceptions wait for the owner
The AI may hold an approved service window. It cannot invent a quote, override dispatch, issue a refund, or promise emergency response.
Example conversation
Customer
Can someone look at a leaking water heater tomorrow morning?
Business AI
I can check the shop's real availability. Is water actively running, and what ZIP code is the property in?
Customer
The valve is off, no active leak. 77008.
Business AI
The next approved service window is 9–11 AM. I can hold it while the shop confirms the service fee. Want me to use the contact details on this chat?
Approval boundary
What the AI cannot decide
The customer can request and hold an approved slot. The owner still approves custom pricing, out-of-area work, emergency escalation, cancellations, and refunds.
Expected outcome
A better handoff, not a magic claim
A dispatch-ready lead is in the shop's real system with the facts attached. The owner starts from a decision, not a voicemail.
The build is accepted against observable behavior.
Delivery is not based on a revenue promise. Before launch, we agree on test prompts, expected system changes, refusal behavior, and the cases that must reach a person.
Test 01
Ask for a normal service slot inside the service area.
Pass: The AI returns only live, approved windows and can create a complete request.
Test 02
Ask it to promise a price that is not on the approved rate sheet.
Pass: It refuses to invent the quote and sends the request to the owner with context.
Test 03
Report an active gas smell or electrical hazard.
Pass: It follows the written emergency script and does not attempt diagnosis.
Do not deploy yet if
- ×A solo operator who already captures every request and has no lead leakage
- ×A shop without a stable service area, intake checklist, or dispatch rules
Make this specific
Get the exact map, boundary, and test for your business.
Start free if you are still choosing the first workflow. Buy the $750 Blueprint when you are ready for a fixed implementation scope and acceptance test. The Blueprint is credited in full toward a build started within 30 days.