HeyHelp comes with a self-learning system that improves its tagging accuracy over time. Here’s how it works and how to review insights.

What counts as "learning"

Tag swaps are what matter

1. If HeyHelp applies a label (e.g. “Action required”) and you swap it for another (e.g. “FYI”), HeyHelp records that as a correction.
2. These corrections are later turned into insights you can review and approve.

No label → adding a HeyHelp label won’t teach it

1. If you have an old email with no HeyHelp label, applying a HeyHelp label manually does not count as training data.
2. Learning only happens from swaps (one HeyHelp tag replaced by another).

Removing labels doesn’t affect learning

1. If you strip a HeyHelp label and leave the email unlabeled, that action isn’t logged as a correction.
2. Don’t worry — removing tags won’t “confuse” the system.

Reviewing and approving insights

To keep you in control, HeyHelp doesn’t automatically retrain itself. Instead, it shows you the patterns it detected from your corrections:

1. Go to your HeyHelp dashboard → Auto tagging.
2. Scroll down to AI Learning Insights.
3. After a few manual corrections, you’ll see AI-generated patterns (e.g. “Emails from *@company.com → tag as FYI”).
4. Review each one and click the X button to ignore it.

This way, you decide what the system learns.

🙋 Still need help?

Drop us an email at hey@heyhelp.ai or a chat and our team will be happy to help.

Updated on: 22/09/2025