Published 1 August 2026 · 5 min read
The legal floor and the sensible standard are not the same height
Businesses usually ask this question hoping for a rule that settles it. There is not one in the Do Not Call provisions, which concern themselves with whether a message is marketing and whether the number is on the registry, not with how the caller identifies itself.
Treating the absence of a rule as permission is the wrong read. The relevant question is not "must we?" but "what happens when the person finds out?" — and on a recorded call to an existing customer, they will.
Why disclosure usually helps the call
The fear is that saying so causes hang-ups. What we see more often is the opposite failure: a caller who suspects something is off spends the first twenty seconds trying to work out what they are talking to instead of listening to the question. Ambiguity costs more attention than disclosure does.
A clear, unfussy introduction also sets the right expectation about what the call can do. Someone who knows they are speaking to an automated system asks the automated system a suitable question. Someone who thinks they have reached the front desk asks about their test results.
How to word it
The introduction that works is short, names the business, states the purpose, and does not make a performance of the technology. Something in the shape of: a courtesy call on behalf of the clinic, about the appointment on Thursday.
Three things to avoid. Do not open with a paragraph about AI — it delays the reason you called and invites the person to engage with the novelty instead of the question. Do not claim a human name and a personal relationship. And do not deflect a direct question: if someone asks whether they are talking to a person, the agent should answer plainly and offer to pass them to one.
What matters more than the disclosure itself
Disclosure is table stakes. The things that actually determine whether an automated call is received well are what happens when it goes off-script.
Does the agent recognise a question it should not answer and hand over cleanly? Does "stop calling me" work first time, permanently, without an argument? Can the person reach a human if they want one? A call that discloses and then traps someone in a script is worse than one that does not disclose and behaves impeccably.
This article is operational guidance, not legal advice. Your organisation is responsible for its own assessment of its obligations and any sector-specific requirements.
Sources
Related questions
What should the agent say if asked "am I talking to a robot?"
It should answer directly that it is an automated assistant calling on behalf of the business, and offer to arrange for a person to call back. Evading the question is the one response that reliably damages trust.
Can the agent use a human-sounding voice?
Yes, and natural-sounding voices make calls easier to follow. Voice quality is a different question from honesty about what the caller is. A natural voice with a clear introduction is fine; a natural voice plus a false claim to be a named staff member is not.
More from Rekindle
- PDPA and Do Not Call rules for automated calls in Singapore
Which automated calls need a Do Not Call check, which are exempt, and the operational questions to settle before a voice agent dials anyone.
- How to reduce no-shows at a Singapore clinic
Reminder timing, the confirmation ask, and why making it easy to cancel reduces no-shows rather than causing them.