No invented facts
Unknown price, policy, availability, or service information should not be guessed.
Risk and operations guide
What an AI receptionist should never be allowed to do
A reliable AI receptionist is defined as much by what it refuses to do as by what it can complete. Clear boundaries reduce bad bookings, false promises, privacy problems, and dangerous advice.
No unsafe advice
Emergency and licensed decisions need approved human or emergency paths.
No fake identity
Callers should know they are speaking with an AI assistant.
No silent failures
If an action does not complete, the caller should not be told that it did.
Do not improvise safety-critical advice
- Medical, legal, financial, or licensed advice
- Electrical, gas, chemical, structural, or equipment safety instructions
- Emergency dispatch or 911 replacement
- Directions that could put a caller in physical danger
Do not invent business decisions
- Unapproved discounts
- Prices that require inspection or judgment
- Appointment availability not returned by the calendar
- Refund, warranty, or exception promises
- Services or geographic areas the company does not offer
Do not collect unnecessary sensitive data
- Full payment-card numbers by voice
- Passwords
- Government identifiers unless a separately reviewed workflow requires them
- Health or other sensitive information without a legitimate need and suitable safeguards
Do not hide that it is AI
A short disclosure sets the right expectation and makes it easier for the caller to ask for a person. Deceptive impersonation creates trust and compliance risk.
Do not claim success when a tool failed
Every booking, payment-link, CRM, or messaging action should have an explicit success result. Timeouts and ambiguous responses need verification or fallback, not optimism.
Common questions
What businesses usually want to know first
Should an AI ever quote prices?
Only when the price can be determined from objective rules you approved.
Should it transfer when the caller asks for a person?
The workflow should support a reasonable human path where available.
Can it handle emergencies?
It can follow approved emergency language and escalation, but it should not replace emergency services or professional judgment.
Keep researching
Related pages that answer the next buying or implementation question.
Buyer comparison
AI receptionist vs hiring another person: decide by the work, not the hype
Compare an AI receptionist with hiring an in-house receptionist across availability, judgment, management, integrations, cost structure, and customer experience.
Read page →Safety guideHow should an AI receptionist handle an emergency call?
Design safer emergency and urgent-call handling with clear detection questions, emergency instructions, human escalation, and no improvised advice.
Read page →Map this around your real calls.
Tell us which calls are being missed, how you schedule today, and where a human still needs to stay in the loop. We will tell you what is realistic before you commit.