Classify each inbound email into urgent_action / awaiting_response / fyi / promotional / spam and send an in-app notification for urgent_action + awaiting_response (subject, sender, suggested next action). Promotional / spam classify silently — no notification noise on bulk mail. Fires automatically on inbound mail; the classifier model sees only PII-aliased text, never the raw sender.

How it works 18 steps

Inspect the data fetches, transforms, gates, and output this recipe runs.

Data fetch

▼ message ?

▼

Process (17 steps)

■ subject default

Apply default

■ from_email default

Apply default

■ body_read ?

■ body default

Apply default

■ body_present compare

Check if body has data

■ skip_empty compare

Check if body present does not equal true

■ protect pii-protect

Apply pii-protect

■ classify ?

skip: step.skip_empty equal true

■ restore pii-restore

Apply pii-restore

■ category default

Apply default

■ confidence default

Apply default

■ is_actionable any

Check if any of the conditions are true

■ below_confidence compare

Check if confidence is less than setting min confidence

■ skip_notify any

Check if any of the conditions are true

■ notify_title switch

Map category to one of: urgent_action, awaiting_response

■ notify ?

skip: step.skip_notify equal true

■ card to_summary

Format results as a summary card

Settings 2 configurable

Configurable at install. Defaults shown — change them anytime in Recued.

channels setting = in_app

min confidence setting = 0.6