What is Email Intent Classification?

When you forward an email, what do you want to happen? Save it for reference? Get an answer to a question? Extract action items? Traditional email-to-notes tools don't ask—they just dump everything into a pile.

Intent classification changes this. AI determines what you want and routes the email accordingly.

The Four Intent Types

Save Note (Default)

Most forwarded content defaults here. The email becomes a searchable note—archived, organized, and retrievable.

Trigger signals: Forwarded content, "FYI," informational language, newsletters

Ask Question

When you send a question, the system searches your knowledge base and generates an answer.

Trigger signals: Question marks, interrogative words (what, when, how, why)

Add Todo

Emails with action items become tasks. Each action item is extracted and linked to the source note.

Trigger signals: "Action items," "TODO," imperative verbs, deadline language

Research

Questions about topics not in your knowledge base trigger external research. Results are compiled and saved.

Trigger signals: "Research:," external topic references, explicit research requests

How Classification Works

Natural Language Processing

AI analyzes both subject line and email body, looking for patterns that indicate intent. It's trained on thousands of examples to recognize common formats.

Confidence Scoring

Each classification comes with a confidence score. High-confidence classifications route immediately. Low-confidence defaults to "Save Note"—the safest option.

Manual Override

If classification is wrong, you can manually recategorize. The system learns from corrections over time.

FAQ

What is email intent classification?

Email intent classification is an AI capability that analyzes forwarded emails to determine what action you want: save as a note, ask a question, create a todo, or trigger research. The system routes each email to the appropriate workflow automatically.

How accurate is intent classification?

Intent classification is highly accurate for clear signals. When the AI has low confidence, it defaults to the safe option (save as note). You can override classifications manually if needed.

For a deeper dive into each intent type and tips for clear signals, read our detailed intent classification guide.