Artificial Intelligence · Evidence and source dates listed below
How do I turn a long message into a task list?
Preserve dependencies and unanswered questions.
Watch the briefing
The short answer
Preserve dependencies and unanswered questions.
Worked example
FICTIONAL EXAMPLE; outputs ILLUSTRATIVE, not model-generated
What this does not establish
Original fictional inputs, mock outputs and editorial diagrams; no personal records, borrowed screenshots or actual model tests.
Full briefing transcript
A long aquarium maintenance email mentions cloudy water, a filter label, a supplier, and a possible visit. A rushed task list says order a filter and book Thursday. Neither action is authorized by the message. Start with three buckets: requests, background, and questions. Then attach the conditions that tell you when a request can actually proceed. A task is not ready simply because it appears on a list.
Our invented email says photograph the filter label by September eighteenth and send it to the maintenance coordinator for review. Do not order a replacement until the coordinator confirms the model. After confirmation, ask the supplier about availability. September twenty fourth is proposed for a visit, not confirmed. Add the visit to the room calendar only after the coordinator accepts its date. Cloudiness and last month's notes are background.
The email also asks whether the required part is already in storage, and requests a stock check with a report to the coordinator. That creates both an unanswered question and a task to answer it. There is no agreed stock check deadline. Ask the draft to retain both, rather than turn an unknown availability into yes or no. Keep the recipient as the owner of direct requests without inventing another person's assignment.
Here is a flawed illustrative list we wrote, not a real tool result. Order a new filter today. Book the visit for September twenty fourth. Cloudy water means the filter needs replacement. The first invents urgency and skips approval. The second confirms a proposed date. The third turns an observation into a diagnosis. None becomes valid because it sounds decisive. We need a corrected administrative draft, not invented aquarium care advice.
The corrected requests begin with the label photograph and its September eighteenth deadline. The supplier enquiry waits for model confirmation; no reply deadline was given. The calendar update waits for an accepted visit date. The stock check stays open until the required part is identified and availability checked. In the questions section, retain the proposed visit date and unanswered storage question. Keep the cloudy display observation and old notes under background.
Check the dependencies as well as the nouns and dates. Can a reader see what must happen before the supplier enquiry or calendar update? Is the only explicit deadline still attached to the photograph? Is a proposed visit still a question? A human coordinator resolves approvals and technical suitability. This lesson organizes the email; it does not diagnose water quality, recommend equipment, send messages, order parts, or change a real calendar.
Preserve dependencies and unanswered questions. Use requests, background, and questions as a practical sorting pattern, then compare each entry with its original sentence. If two buckets overlap, explain why, as with the stock question and the task to answer it. The goal is a reviewable list, not premature certainty. Next, we will compare two venue offers using the same fields, even when one leaves an important price blank.
One insight you can use
Preserve dependencies and unanswered questions.
Disclosures
AI-assisted production and synthetic narration. Original teaching examples and diagrams; linked third-party sources retain their respective rights.
Original sources and limits
- developers.openai.comChecked 2026-09-17
Explicit instructions, context, examples and non-deterministic model output; developer guidance adapted to human-reviewed fictional drafts.
- developers.openai.comChecked 2026-09-17
Define evaluation criteria and use human feedback; developer guidance, not evidence that this fictional workflow saves time.