How to make sure your AI never calls someone who said stop
John Nguyen — Founder of Patio, licensed Texas REALTOR®
A man told Keisha's phone AI, "Stop calling me." Three days later, it called him again. It took Keisha a week to find out why.
One of the 9 agents at her real estate office loaded an old list with his number.
Calling people comes with more rules than most other work a business does. Consent means a person said yes to your calls. An opt-out means they said stop.
Before you buy a phone AI, ask the company 5 questions, and walk away if it dodges.
Can it answer calls without making them?
When someone calls Keisha's office, they chose to reach her, so the rules are light. The AI should say it is an AI. It should get them to a person when they ask.
Calls to people who did not reach out first are not the same. Those come with rules on consent, do-not-call lists, the time of day and saying who is calling. Each call that breaks them is its own risk.
So ask: can I get my calls answered without a machine that dials people? If not, you must follow the hard rules for calling people to get your phone answered.
Where does it keep each person's yes?
For any call going out, you need proof of the yes that lets you call. It shows who said yes, when, how, and to what. A good one looks like this: "Ana, May 2, our website, calls about homes."
It should sit with that person's name.
A bad company says the yes is yours to track. Then it calls people without checking. Walk away from that answer.
What happens when someone says stop?
A person should be able to say "Stop calling me" on the call, with no email or website. Then 4 things should happen:
- The AI notes the stop right away, next to that person's name.
- The stop works before the next call, not the next morning.
- It holds for the whole office, not just one agent.
- It still holds when someone loads the same list again.
When someone tells your office to stop, they expect the whole office to stop. A bad answer is "The agent marks them as do-not-call." That depends on a person, and someone will forget.
Which do-not-call lists does it check, and when?
A do-not-call list is like a "Please don't knock" sign for phones. You must check 3: the U.S. list, any state list that applies, and your own. Your own list is the one people miss most.
Ask if it checks the list at the moment of each call. A list checked 3 days before the call is 3 days old.
Can it prove all of this later?
Complaints arrive months after the call. Then Keisha must show what happened on that one call. What did the AI say, which yes let her call, and had the person said stop?
Some states need every person's consent to record a call. If the AI records, ask what it does about that. Ask where it keeps saved calls, for how long, and who can hear them.
Look at Jessica, the AI in Patio, an app for real estate offices. She answers each call as it comes in. She tells each caller she may record it.
But telling someone is not the same as getting their yes. So still ask how each AI gets a yes and keeps proof.
What happens if you skip this
Say Keisha skips these questions. An agent loads an old list again, and the AI calls 40 people who said stop. Each call goes out in her office's name, so her office gets the blame.
The short version
Before an AI makes calls for you, ask the company 5 questions. A good one shows you where it keeps each yes and how one stop reaches the whole office. If it dodges a question or calls the yes your job, walk away.
This is not legal advice, the kind a lawyer gives about your own case. Phone rules come from U.S., state and local law, and they change. Have a lawyer check how you call people.
Keep reading
Which jobs to give AI, and which to keep
AI can pick up your calls, write first drafts and remind you to follow up. Keep people, prices and contracts for yourself, and check its work.
How to check AI listing ads for Fair Housing
AI writes home ads that sound nice but can still break Fair Housing law. Ask it for facts only, then check each sentence for the building or a person.