How long does InboxIQ take to learn my style?
How long does InboxIQ take to learn my style?
InboxIQ's AI starts working from the first email — it does not require a training period. However, it does improve over time as you provide feedback.
Day one
On day one, InboxIQ uses its general-purpose AI model to classify emails and draft replies. The drafts will be correct in structure and tone, but may not perfectly match your specific voice or the conventions of your product.
Getting better over time
InboxIQ learns from two types of feedback:
1. Label corrections If InboxIQ puts an email in the wrong category, you can correct it by changing the Gmail label or Outlook category. InboxIQ detects label corrections automatically and uses them to improve future classifications.
2. Knowledge base content The more context you add to your Knowledge Base (past replies, help articles, product docs), the better InboxIQ understands your product and the more accurate its draft replies become.
Timeline
| When | What to expect |
|---|---|
| Days 1–3 | Good classifications; drafts may lack product-specific detail |
| Week 1–2 | Noticeably more accurate as corrections feed back |
| Month 1+ | Drafts closely match your style with sufficient KB content |
The exact timeline depends on email volume and how much KB content you've provided.
Speed it up
- Add your most-referenced help articles and past replies to the Knowledge Base
- Correct misclassified emails promptly — each correction is a training signal
- The higher your email volume, the faster the feedback loop
Related articles
Still need help? Contact support →