Docs/AI agentsBest practices
Managing opt-outs and STOP requests
When a lead asks to be left alone, the AI stops. It closes the conversation politely and then turns itself off for that contact.
On this page
What happens when someone opts out
| Step | What Sendify does |
|---|---|
| The lead asks to stop | The AI recognizes it, whether it is the word STOP or a plain language request |
| Acknowledgement | The AI normally sends one short, polite closing message |
| The AI turns itself off | No further AI replies on that conversation |
| Follow ups | Stop with it. Nothing further is scheduled for that contact |
Why the polite close matters
Plain language counts
Because the AI is reading the conversation rather than matching a keyword list, "don't text me", "take me off your list", and "not interested, stop messaging" all land the same way as STOP. You do not maintain a phrase list.
This is also why we keep A2P style opt-out wording out of your copy in the first place. That reasoning is in the iMessage guide.
GoHighLevel accounts
Sendify stopping the AI is separate from your CRM marking the contact. If you want the contact set to DND in GoHighLevel, create a GoHighLevel automation that does it. That keeps your CRM the source of truth for contactability, which is how we recommend running it anyway, see what syncs between GoHighLevel and Sendify.
Sendify Direct and API accounts
The same behavior applies: the AI closes the conversation and turns itself off. Your own system stays the source of truth for suppression, so mirror the opt-out on your side when you receive the AI status change event. Event details are in the API reference, and you can stop the AI for a contact yourself with the stop endpoint.
Test it before you launch
Opt-out checks
- Reply STOP from a test phone and confirm the AI acknowledges then goes quiet
- Reply with a plain language request and confirm the same result
- Confirm no follow up arrives afterwards
- If you built a GoHighLevel DND automation, confirm it fired
Add these to your launch checklist run.
Opting someone back in, or out, by hand
An opt-out is not permanent. Staff can opt a contact back in manually when the person asks to hear from you again, and staff can also opt a contact out by hand, for example when someone asks over the phone or by email rather than by text.
Until someone is opted back in, the AI stays off for them and nothing is scheduled or sent.
Your compliance responsibility
You own consent and opt-out handling
FAQ
What happens when someone replies STOP?
The AI stops and turns itself off for that contact. It normally sends one short, polite acknowledgement first, then it is done: no further AI replies and no further follow ups.
Does it only recognize the word STOP?
No. The AI is reading the conversation, so a plain language request like 'don't text me' or 'take me off this list' is treated as an opt-out the same way.
Why does it reply at all before stopping?
A short courteous close reads like a person and reduces the chance of the contact reporting the number. Silence after an angry message is more likely to get a line flagged.
Do follow-ups keep going?
No. When the AI turns itself off for a contact, the timed follow ups stop with it.
Does the contact get marked as do not contact in GoHighLevel?
Not on its own. If you want DND set in GoHighLevel, build a GoHighLevel automation that sets it. Sendify stopping the AI and GoHighLevel marking DND are two separate things.
How do I test this before launch?
Use your own phone as a test lead, start a conversation, reply STOP, and confirm the AI acknowledges and then goes quiet. Repeat with a plain language request like 'please don't text me again'.
Still need a hand?
If your volume is going to run past the daily or hourly limits, or something is not behaving the way these docs describe, talk to our team.