How InboxIQ learns from corrections
How InboxIQ learns from corrections
InboxIQ improves automatically as you use it. Every time you correct a classification, InboxIQ records it and adjusts future behaviour.
How corrections work
When InboxIQ misclassifies an email:
- Gmail — move the email to the correct
InboxIQ/label in the sidebar. InboxIQ detects the move on the next poll. - Outlook — right-click → Categorize → select the correct
InboxIQ/category.
No forms to fill in. No buttons to click. The correction happens through your normal inbox workflow.
What InboxIQ learns
Each correction updates two things:
-
Sender profile — InboxIQ builds a profile for each sender domain. After a few corrections from the same domain, it classifies future emails from that domain more accurately.
-
DSPy training data — Corrections are collected and used to periodically retrain the underlying classification model. This improves accuracy across all emails, not just from corrected senders.
How long before improvements show?
- Sender profiles — you'll notice improvement after 3–5 corrections from the same sender domain, often within a few days.
- Model retraining — happens periodically (not in real-time). Improvements from retraining are typically visible after 2–4 weeks of regular use.
Related articles
Was this article helpful?
Still need help? Contact support →