AI Receptionist Call Flow Templates for Service Businesses

A good AI receptionist call flow identifies why someone called and collects only what you need. It finishes one task and has a defined route to a person. The five templates below give you a starting point for a service business.

These templates are part of our AI Front Desk guide. They are our recommendations, not an industry standard.

Platform details were checked against vendor documentation on 9 November 2026. This is general information, not legal or medical advice.

The skeleton every flow shares

Use these steps as the frame for every template.

1. Greet. Name the business and say what is answering.
2. Identify the reason. Ask one open question.
3. Check for urgency. Listen for your urgent keywords at any point.
4. Do the one job. Book, take a message or answer the question.
5. Confirm. Read back names, numbers and times.
6. Close. Say what happens next and when.

Add one rule across all steps. Any request for a person moves to a transfer, without argument.

Prompt or structured flow?

Retell offers two approaches. Its conversation flow overview says: "Conversation flow agents divide a task into nodes, each with its own instructions or actions."

It also says: "For your first version, we recommend starting with a single prompt agent unless you need the additional structure." The same page warns that node boundaries can make conversations harder to handle. Examples include a caller who shares information early or changes their mind.

So start simply, and add structure where testing shows you need it. Retell's flow best practices also advise: "Add handling for failures or exceptions only when the need is obvious or supported by evidence." A failed transfer is an obvious need, so build that one from the start.

Template 1: New customer inquiry

Goal: a booked appointment or a qualified message.

1. Greet and ask how you can help.
2. Confirm the service is one you offer.
3. Confirm the address or area is one you serve.
4. Collect name, callback number and a short job description.
5. Offer available times, or take a preferred time.
6. Read back the details and say what happens next.

If it fails: when the service or area doesn't fit, say so politely and end the call. If the calendar lookup fails, take a message and give a callback time.

Template 2: Change or cancel an appointment

Goal: an updated appointment, or a message for staff.

1. Ask for the name and phone number on the booking.
2. Confirm the existing appointment before changing anything.
3. State your cancellation or rescheduling policy from your written rules.
4. Offer new times, or confirm the cancellation.
5. Read back the result.

If it fails: when the booking can't be found or the policy needs an exception, transfer to staff or take a message. Don't let the AI make exceptions.

Template 3: Urgent call

Goal: the caller reaches a person, or emergency services, quickly.

1. On any urgent keyword, stop the current step.
2. If someone is in danger, tell the caller to hang up and dial 911.
3. For an urgent job, say you are connecting them to someone now.
4. Transfer to the on-call number.
5. If nobody answers, take the callback number and alert the on-call person by text.
6. Tell the caller what will happen next, with no promise you can't keep.

Products handle this in different ways. Jobber's AI Receptionist page says you can choose urgent keywords, and a match can transfer the caller or send you a text alert.

Do not let the AI assess medical or mental health emergencies. For a mental health crisis, the 988 Lifeline says: "The 988 Lifeline is available 24/7/365."

Template 4: After-hours message

Goal: a complete message, ready for morning.

1. Say the office is closed and when it reopens.
2. Check for urgency and send urgent calls to Template 3.
3. Collect name, callback number and the reason for the call.
4. Offer a booking, if your system allows it.
5. Say when someone will respond.

Jobber's page says you can route calls to its receptionist after hours or after a set number of rings. Our after-hours options guide compares other approaches.

Template 5: Spam, sales and wrong numbers

Goal: a short, polite exit that doesn't take up staff time.

1. Ask how you can help, as with any caller.
2. If it's a sales pitch, say you'll take a message and collect a company name and number.
3. If it's a wrong number, give the business name and end the call.
4. Log the call so your team can review it.

Don't let this path block real customers. Review a sample of calls marked as spam, so you can catch anyone who was filtered by mistake.

Build the transfer properly

The transfer is the part most worth testing. Retell's call transfer documentation says a transfer node moves the call to another phone number or SIP address. It lists cold transfers, and warm transfers where the agent can first try to detect whether a human answered, if that setting is turned on.

Retell's default wait for that detection is 30 seconds, and the node has its own path for failed transfers. It also says: "This node only works during phone calls instead of web calls."

Plans differ too. Dialzara's pricing page lists blind and warm transfers on its Business Elite and Done for You plans, and not on its lower plans. Our guide to when an AI should hand off to a person covers the triggers.

Where disclosure fits

The greeting is where you may need to say the caller is speaking with AI, or that the call is recorded. Requirements vary by place and by industry, so we are not giving you legal wording here. See our AI Front Desk guide and have a qualified attorney review your greeting.

How to test a flow

Run each template by calling it from a real phone.

• Give details in the wrong order, or all at once.
• Change your mind halfway through.
• Ask for a person during each step.
• Say an urgent keyword in the middle of a booking.
• Have the transfer number go unanswered.

Write down each failure and fix the cause. Then repeat the test, including the paths that worked before.

Next step

Choose the template that matches your most common call, and map it on paper with your staff. If your flow needs custom connections to your own systems, our AI agent development team can scope it with you.

Frequently asked questions