Pattern Learning
Pattern learning lets MailPrism watch how you organize your inbox — what you archive, label, star, and delete — and spot repeating habits. Those patterns power rule suggestions, so you can automate the things you already do by hand.
You're in full control: you choose which actions are watched, and you can turn the whole feature off. Settings live in Settings → Privacy (the Pattern Learning card).
What MailPrism learns from
When pattern learning is on, you decide which of your actions it tracks:
| Action watched | What it tracks | On by default? |
|---|---|---|
| Archives | When you archive emails | Yes |
| Labels | When you apply or remove labels | Yes |
| Stars | When you star or unstar emails | Yes |
| Deletes | When you delete emails | No |
Deleting is a blunt action — people delete for many different reasons — so MailPrism treats it as optional and leaves it off unless you opt in.
Each toggle is independent, so you can let MailPrism learn from labels and archives while ignoring everything else.
What it powers
Detected patterns feed MailPrism's suggestions: instead of building every rule from scratch, you get proposed automations based on what you actually do. Pattern learning is treated as a core feature of how MailPrism gets more helpful over time.
You can also see how much it has learned. The Privacy page shows running totals for:
- Patterns Detected — distinct habits MailPrism has spotted.
- Actions Logged — email actions recorded for learning.
- Learning Corrections — times your feedback adjusted what the AI learned.
Correction history
When you give feedback on the AI's classifications — agreeing, disagreeing, or correcting them — MailPrism records it so the AI can learn and improve its accuracy over time. You can review this history in Settings → Privacy → Classification Feedback, where each entry is tracked as:
- Agreed — you confirmed the AI got it right.
- Corrections — you changed the AI's result.
- AI Learned — feedback the AI has since incorporated.
From that page you can filter your feedback, export it, mark items as learned, or clear it.
Turning pattern learning on or off
In Settings → Privacy → Pattern Learning, the Enable Pattern Learning toggle controls the whole feature. Turn it off and MailPrism stops watching your actions entirely; the per-action toggles only apply while it's on.
The Privacy page also lets you export or clear your learning data at any time, and set how long MailPrism retains it. See AI privacy & consent.
→ Related: AI, Explained · Smart rules