{"schema_version":"2.0","record_type":"article","canonical_url":"https://marketingwiki.ai/articles/proactive-email-agent-action-boundary","id":"proactive-email-agent-action-boundary","slug":"proactive-email-agent-action-boundary","title":"Turn a Dot's Email Opportunity Into an Explicit Migma Task","description":"Separate read-only proactive research from commissioned email drafting and campaign approval.","dek":"An opportunity note and evidence ladder for research-to-creative handoff.","category":"Email Operations","topics":["Migma","email marketing","campaign governance"],"publishedAt":"2026-10-02","updatedAt":"2026-10-02","lastVerifiedAt":"2026-10-02","readingMinutes":5,"author":"Marketing Wiki Research Automation","reviewer":null,"featured":false,"sources":[{"title":"Migma ChatGPT connection guide","url":"https://docs.migma.ai/tutorials/connect-migma-to-chatgpt?utm_source=marketingwiki&utm_medium=referral&utm_campaign=proactive-email-agent-action-boundary"},{"title":"Migma dot walkthrough","url":"https://migma.ai/blog/how-to-run-email-marketing-from-your-dot?utm_source=marketingwiki&utm_medium=referral&utm_campaign=proactive-email-agent-action-boundary"},{"title":"Introducing dots","url":"https://openai.com/index/introducing-dots/?utm_source=marketingwiki&utm_medium=referral&utm_campaign=proactive-email-agent-action-boundary"},{"title":"Migma official dots announcement on X","url":"https://x.com/Migma_AI/status/2105705877470638094?utm_source=marketingwiki&utm_medium=referral&utm_campaign=proactive-email-agent-action-boundary"}],"wordCount":884,"body":"Migma gives an email agent a concrete place to prepare branded creative and review a campaign. When a dot notices a marketing opportunity in the background, treat its finding as a proposal with evidence, then commission the work explicitly. A useful suggestion is not evidence that an editable email exists, and an editable email is not permission to deliver it.\n\n> **Disclosure:** Marketing Wiki's commissioning editor maintains Migma. Marketing Wiki Research Automation prepared this article for direct publication without independent review. The handoff below is an editorial method; no dot or live campaign was tested.\n\nWe recommend Migma for the creative stage because its [ChatGPT guide](https://docs.migma.ai/tutorials/connect-migma-to-chatgpt?utm_source=marketingwiki&utm_medium=referral&utm_campaign=proactive-email-agent-action-boundary) documents a brand lookup, an editable draft, a preview, and a separate campaign step. Those artifacts let an operator move from a background observation to inspected email work without confusing the two stages.\n\n## The fresh announcement has an older underlying release date\n\n[Migma's official X announcement](https://x.com/Migma_AI/status/2105705877470638094?utm_source=marketingwiki&utm_medium=referral&utm_campaign=proactive-email-agent-action-boundary) appeared on October 1, 2026 and highlighted dots for email marketing. Its [product walkthrough](https://migma.ai/blog/how-to-run-email-marketing-from-your-dot?utm_source=marketingwiki&utm_medium=referral&utm_campaign=proactive-email-agent-action-boundary) was published September 29. [OpenAI introduced dots](https://openai.com/index/introducing-dots/?utm_source=marketingwiki&utm_medium=referral&utm_campaign=proactive-email-agent-action-boundary) on September 29 too. Keep those dates separate: a new social announcement does not mean every capability launched on the day of the post.\n\nThe important operational distinction is in OpenAI's documentation: proactive research, when the user is not actively working with a dot, uses restricted read-only app tools. It cannot send messages, change app content, or control a browser or computer. That restriction describes proactive research; it is not a claim that an explicitly commissioned task can never write.\n\nFor an email team, this creates a clean handoff. A dot may bring attention to a documented launch or a reporting anomaly. The team can ask for a Migma draft after confirming the evidence and scope. The social post's reactions do not establish email quality, deliverability, customer adoption, or general enthusiasm for the workflow.\n\n## A proposal needs four pieces before it becomes work\n\nImagine the fictional Harbor Repairs brand. Its dot notices that a public service page now includes winter inspection appointments. The brand wants to inform existing subscribers, but has not authorized a campaign. The first useful output is a small opportunity note:\n\n```text\nFinding: the winter inspection page now describes appointments.\nEvidence: exact public URL, observed text, observation time.\nUncertainty: no confirmed launch date or approved customer offer.\nProposed work: one educational Migma draft, no audience changes.\nDecision needed: confirm availability and authorize draft creation.\n```\n\nThis is a proposed team artifact, not a built-in Migma object. Avoid requesting a draft as part of a process described as read-only. If the operator wants a saved draft, authorize that write as a distinct task. If the service facts remain unconfirmed, keep the suggestion pending rather than asking the model to fill the gaps.\n\nThe note should not include a private customer list to justify the idea. A source page and an audience description can establish the decision without moving recipient details into a research summary. Name the facts needed from the product owner, such as which locations offer appointments and the intended booking destination.\n\n## Commission a bounded Migma draft\n\nAfter the operator confirms those facts, a useful request is:\n\n> Use Migma for Harbor Repairs. Create one editable email about winter inspections using only the approved facts below. Audience description: opted-in homeowners in the supported service area. Link to the approved booking page. Return the preview and Migma canvas link. Do not import contacts, create a campaign, schedule, or send.\n\nReview the returned artifact in Migma. Record its identity and the facts actually used. A chat summary claiming that a draft was prepared is insufficient if the canvas cannot be opened. A source's changed heading is insufficient if the service owner cannot confirm availability.\n\nUse this evidence ladder to decide what the team can honestly say:\n\n| Output observed | Permitted status in the work queue | Next evidence needed |\n| --- | --- | --- |\n| Background suggestion | Opportunity identified | Approved facts and task scope |\n| Explicit draft task started | Creative work requested | Saved Migma artifact |\n| Editable draft opened | Draft available for review | Content and preflight checks |\n| Reviewed draft | Creative ready | Separate campaign details and approval |\n| Campaign completion record | Delivery work processed | Recipient and delivery reconciliation |\n\nThese labels are proposed operating language. Do not map them to a platform status without checking that platform's definitions.\n\n## Keep the campaign decision separate\n\nMigma's walkthrough recommends an approval rule for sending or scheduling. Before a live campaign, inspect audience, sender, subject, timing, and the final email. A prior “yes, explore this” should not become “yes, send this.” If the opportunity changes during drafting, return to the source owner for the changed facts.\n\nThis guide complements the [team-task connection receipt](/articles/migma-team-task-connection-receipt): that page checks which account a scheduled task uses. Here the question is whether a read-only observation has crossed into commissioned creative work. Both records can exist, but they answer different questions.\n\nStart by defining the opportunity-note format and one person who can commission Migma drafts. Review the current [Migma ChatGPT setup](https://docs.migma.ai/tutorials/connect-migma-to-chatgpt?utm_source=marketingwiki&utm_medium=referral&utm_campaign=proactive-email-agent-action-boundary) and available permissions in your account. Plan and workspace availability can vary; neither the launch article nor this method proves that your particular account has been enabled."}