About this article
This article was created using an automated generation workflow powered by generative AI. Official Microsoft documentation was reviewed on October 20, 2026, and a prompt along with verification steps using dummy data was created. Execution in the Copilot interface has not been performed.Verification Status: 📘 Official Information Confirmed – Prompt Execution Unconfirmed
- Supported Environments, Capabilities, and Limitations (Official Confirmation: October 20, 2026)
- Organizing scattered work notes into four columns
- Verifying verbs in the achievement column
- Reasons for using the same fact in multiple fields
- Change only one place and verify whether the status shifts
- Formatting into a shape that enables the reader to make decisions
- When keeping numerical values, also include the population size and conditions.
- For continuous use, retain the delta from the previous week.
- Official and Primary Information
Supported Environments, Capabilities, and Limitations (Official Confirmation: October 20, 2026)
| Item | Target and Scope of this Example |
|---|---|
| Product / Entry Point | Microsoft Copilot Chat for work (formerly Microsoft 365 Copilot Chat).Assuming standard chat on the web versionAssuming standard chat on the web version |
| Account and License | Work account authorized for organizational use. Check your license and administrator settings to confirm availability. Personal Copilot is out of scope for this example. |
| Additional Contracts and Billing | Additional features within Office apps and pay-as-you-go agents are not used in this example. Do not interpret this as enabling features that require additional contracts. |
| Input and Storage Location | Directly paste fictional text. Do not connect local paths, OneDrive, SharePoint, or the inbox. |
| Actions to Try | Referencing, organizing, comparing, and drafting inputted text. Content is to be manually verified against the original text by a human. |
| Out of scope for this example | File search, direct document editing, moving, and deletion, sending emails, and registering in enterprise systems |
| Limits and availability | Do not assume preview functionality. Numeric values and physical builds for input length, frequency, region, and screen differences are unverified. Split long inputs and verify availability on the screen. |
| Information management | Use only fictitious data initially. For expansion to business information, verify organizational permissions and information classification. Do not rely solely on prompt instructions as a guarantee for access control or disabling web search. |
Intending to make weekly reports more readable can sometimes make unstarted work look completed. "Conducted testing" and "passed testing" are different, and "contacted the client" and "agreed with the client" are also different. When formatting weekly reports with Copilot Chat, provide conditions that prevent state changes before shortening the text.
Organizing scattered work notes into four columns
The following is a draft note for a fictitious internal asset search tool. Do not use actual business logs or user information. Paste the prompt and notes together into a new conversation. This does not include operations for automatically searching past weekly reports or project documents.
次のメモだけで週報の下書きを作ってください。 区分は「今週の実績」「未完了・課題」「来週の予定」「判断が必要なこと」。 完了・実施・確認・合意を同義にしないでください。 進捗率、効果時間、担当、期限は入力にない値を補わないでください。 各項目の末尾に根拠行を付けてください。 最後に、不足情報を最大3点示してください。 L1：検索画面の試作を作成した。実際の利用者による確認はまだ。 L2：ダミーデータ20件で検索を実行した。2件は表記ゆれで見つからなかった。 L3：表記ゆれ対応の方法を来週比較する予定。 L4：運用担当へ保管場所の確認メールを送った。返事はまだない。 L5：公開日は決めていない。利用者確認の後に相談する。
Expected achievements are screen prototyping, executing searches, and sending confirmation emails. User verification, handling terminology variations, agreeing on storage locations, and deciding release dates are excluded. If outputs like "tool development completed" or "all 20 items successfully searched" appear, the phrasing has become stronger than the facts.
Verifying verbs in the achievement column
Misunderstandings in weekly reports often appear in verbs rather than nouns. This is because "checked" can easily conflate different meanings such as asked a question, viewed a screen, or obtained the other party's approval. If the input is only up to sending an email, limit the output to sending as well.
| Note status | Examples of usable expressions | Expressions not usable in this example |
|---|---|---|
| Created prototype | Prototype creation completed | User acceptance completed |
| Executed search | Tested with 20 items, 2 items not found | Search test passed |
| Sent confirmation email | Inquiry submitted, awaiting response | Storage location determined |
| Compare next week | Comparison scheduled | Method determined |
The suffix "completed" is particularly convenient, but you need to be specific about what has been finished. Because "tested" does not show the results, a description like "tested with 20 dummy items, 2 undetected" is more useful for decision-making. Leave enough information for the reader to determine the next step.
Do not let the system automatically calculate progress rates either. Even if three out of five notes contain "done," that does not mean 60% completion. This is because tasks vary in weight, and there may be unwritten processes remaining. If you want to show progress with numbers, determine the planned processes and completion criteria separately before calculating.
Reasons for using the same fact in multiple fields
Conducting the search is a completed achievement, and the two undetected items are issues. Even if the same L2 appears in two categories, it is not necessarily simple duplication. By separating what was done from what remains, readers can simultaneously grasp the progress of the prototype and the remaining tasks.
On the other hand, there is no need to include the exact same text twice. Keep the trial conditions under achievements, and the undetected status under issues. By adding an instruction to Copilot such as "You may use the same basis, but write only the parts necessary for the purpose of each field," you can convey the intent behind the organization.
We also check the difference between risks and problems. The fact that undetected items have already occurred is a problem. While a delay in user confirmation pushing back the release date is a potential risk, you must not add impacts not present in the input as facts. When including inferences, separate them as hypotheses or items to be verified.
Change only one place and verify whether the status shifts
Replace L4 with "Received a response from the operations team and agreed to use shared folder A as the storage location," and test the same instruction in a different conversation. The expected change is that the pending response is resolved and the agreement on the storage location moves to achievements.
However, agreeing on something is different from actually moving the files. The new L4 still does not state that the migration was executed. If the output becomes "Migration to shared folder completed," the status has advanced by one step. Confirming such minor differences helps prevent false progress in lengthy weekly reports.
When making corrections with additional conversational instructions, quote the sentence to be corrected and state, "This description only goes up to agreement and should not be marked as completed." Saying simply "make it more accurate" leaves it ambiguous which boundary should be reverted. Pointing out the problematic verb makes it easier to verify the correction results.
Formatting into a shape that enables the reader to make decisions
For a weekly report aimed at managers, reasons for blockages and decision items may be prioritized over minor operations. For a handover between assignees, reproduction conditions and remaining tasks should be detailed. After confirming the facts, specify the reader and simply change the arrangement of the same content to make it easier to handle.
It is also important not to force the "items requiring decision" field to be filled. This memo does not necessarily imply an immediate request for approval. The release date is scheduled to be discussed after user confirmation, and it may not yet be the stage to have the schedule approved immediately. Fields that are not applicable may be left blank.
When shortening text, avoid removing conditions first. Terms like "dummy data", "user unverified", and "awaiting response" are not decorative; they are information that indicates the limits of the results. For instructions to reduce character count, add a note to "retain state and constraints" and compare the text before and after deletion.
When keeping numerical values, also include the population size and conditions.
A mere count such as "two items not found" does not indicate whether twenty items or two thousand items were tested. Even when shortening a weekly report, retain the population size required for decision-making. In this example, maintaining the condition of twenty dummy data items prevents misreading the results as having been verified against all actual users.
If the purpose of the prototype is to investigate spelling variations, detailing the two undetected items is more valuable than detailing the eighteen matching items. Adding a single sentence explaining what was being tested helps convey the meaning of the numbers. Avoid adding words that evaluate the results as "high precision" based solely on the count.
In risk descriptions, separate occurred facts from future impacts. "Awaiting response" is a fact, but "publication will be delayed by one month because there is no response" cannot be derived from the input alone. If impacts must be written, the person in charge should check the dependent processes and required deadlines, and indicate the premise for the estimation.
When additional questions come from the reader, feed the answers back into the next week's input. Do not expect the AI to know them from previous conversations; explicitly state them in the notes that serve as the basis for the weekly report. Keeping a single official report helps reduce inconsistencies in status across conversations.
Decisions to omit descriptions should also be based on evidence. Distinguish whether an item is omitted because it is minor or because it has been resolved, and keep its status in the management list. Do not treat disappearance from the report text as completion of the work itself.
For continuous use, retain the delta from the previous week.
When creating weekly reports on a regular basis, assigning IDs to incomplete items from the previous week is useful. Compare them with this week's notes to distinguish between completed, ongoing, canceled, and no information. To prevent items from being treated as completed simply because they disappeared from the description, retain "no information" as a distinct state.
Limit inputs provided to the AI to the scope permitted by the organization. Prototype notes may also contain internal names, failure details, or personal evaluations. Do not assume that a short weekly report is safe to publish; select information according to the audience and the input destination.
Evaluation focuses on both readability and fact retention. Compare the number of completed items, pending deadlines, and the basis of numerical values before and after revision. Even if the phrasing is polished, if incomplete items become obscured, the weekly report weakens as a basis for decision-making.
The value of a weekly report lies not in making the work look impressive, but in sharing the conditions for moving forward. By separately describing execution and acceptance, inquiry and agreement, and agreement and implementation, readers can reduce the need for additional confirmations. Leave text formatting to Copilot, and make decisions to advance the status based on evidence.
Official and Primary Information
Product scope and input methods refer to official documentation. Fictional data, prompts, expected examples, and checklists are verification proposals for this article and are not output examples guaranteed by Microsoft.
