AI Email Task Workflow in Mailbase

Overview

Mailbase AI tasks are persisted records, not local-only reminders. Teammates queue tasks from replies, campaign work, analytics, and empty states; MCP clients read them with workspace-scoped tools, complete them with structured results, and the app keeps a review history.

Queue And Complete

The MCP server exposes task tools such as list_ai_tasks and complete_ai_task. OAuth-scoped clients stay pinned to the authorized workspace, and legacy sessions do not guess a workspace when one cannot be resolved.

Completed task payloads are normalized into result types such as draft reply, generated variants, report, classification, and contact enrichment. That keeps the UI predictable even when different AI clients complete the work.

Persistent AI task records
Workspace-scoped MCP task tools
Structured completion result schemas
Completed-task history
No automatic send on completion

Review And Apply

Operators can approve, reject, or apply completed task outputs. Apply actions can draft replies, create templates, add tags, or write contact notes depending on the result type.

Workspace AI policy settings define tone, signature, forbidden claims, and pricing rules so AI clients can pick up local guidance before they write.

Reply draft apply action
Template apply action
Tag and contact-note actions
Review metadata
Workspace AI policy

Common Mistakes

Choosing a tool before deciding who owns deliverability.
Treating DNS authentication as a one-time checkbox instead of an operating baseline.
Mixing product-critical transactional email with experimental marketing sends, with no clear boundary.
Trusting headline metrics (like open rate) that privacy proxies now inflate.