Repeat a public-source investigation and deliver a private draft digest.
Conversation & research · Changing · Documentation checked 2026-10-04
ChatGPT Chat and Work; local Codex and the OpenAI API are separate surfaces.
Tools, usage, and controls vary by plan, client, region, rollout, and workspace policy.
An illustrative recipe. This website does not run the automation.
The goal and the right kind of system
Repeat a public-source investigation and deliver a private draft digest.
A configured scheduled task fits a stable research brief. An agent is only useful if the investigation needs variable intermediate searches; outbound publication should remain outside the task.
Inputs and connected services
Inputs: exact topic, primary-source preferences, time zone, cadence, and review criteria.
Services: search and scheduled-task execution. Connected services are optional; local-folder dependencies do not belong in a cloud-only task.
The steps
An illustrative sequence to adapt and test before using real data or enabling ongoing work.
- Test the research prompt manually and check the sources.
- Create the supported schedule using cloud-accessible sources.
- Inspect the first run’s citations and evidence / delivery dates.
- Revise or pause the task when the purpose or source scope changes.
Outputs and human checkpoints
Outputs: a dated digest, evidence links, and a missing-information note.
Human checkpoints: confirm the schedule and review the first digest; approve any later sharing individually.
Exceptions, permissions, and stopping conditions
Permission boundary: recurring reads and private output only. Provider limits or missing sources produce a visible exception, not invented changes.
Stop / pause on repeated unsupported claims, inaccessible required context, exceeded budget, or expired purpose. A fresh run must restate essential criteria.
Adaptable prompt
Weekly at [time and time zone], investigate [topic] from public primary sources. Return event dates, URLs, evidence type, and uncertainties. Mark no supported change explicitly. Draft for my review only. Do not message people or publish.
Related reading
Search and deep research
Find current information and build a report whose sources can be checked.
Scheduled tasks and event triggers
Repeat a proven task with an explicit schedule, input scope, and review destination.
Permissions, administration, and product boundaries
Audit workspace access, runtime permissions, and external actions before expanding an agent task.
Designing an AI workflow
Make a useful task repeatable with clear inputs, steps, outputs, exceptions, and human checkpoints.
Permissions and meaningful approvals
Decide what a system may read, what it may change, and when someone needs to authorize a concrete action.
Sources and verification
Documentation checked 2026-10-04. Examples are illustrative unless described otherwise.
Web search and deep research · OpenAI
Scheduled tasks · OpenAI
Permissions and approval policies · OpenAI
Workspace administration · OpenAI
Get started with Work · OpenAI