About This Article
This article is created using an automated generation workflow powered by generative AI. Cross-referencing official Microsoft help and update information as of October 27, 2026, it organizes operating procedures, contract terms, and troubleshooting decisions for beginners to verify using actual screens.Actual button operations have not been performed in Microsoft Copilot. The result examples in the text are for practice purposes only.

Verification Status: 📘 Confirmed with official Microsoft specifications – User environment live testing and WordPress display are unverified.

Confirmed on October 27, 2026.Features and interfaces for Copilot vary by personal/work account, free/paid tier, region, and rollout status. Please check the linked official Microsoft information for the latest pricing, limits, and availability conditions.

First: Supported Environments, Capabilities, and Limitations

Understanding the big picture of Copilot Chat: An introductory guide to mastering screens, Smart mode, and the differences between free and paid versions.

When you first open Copilot, you will see a message input field in the center, and conversations, libraries, tasks, projects, discover, Imagine, experiments, and other items lined up on the left. Sometimes "Smart" is displayed below the input field. The more features available, the harder it is to know where to start. Therefore, rather than memorizing all the feature names, this guide connects choosing an entry point based on the task you want to accomplish, understanding screen meanings, recognizing contract differences, and conducting initial practice into a single flow.

For personal practice,Microsoft Copilotopen the application, sign in with your personal account if necessary, click [New chat], and enter the fictitious prompt from Section 3 to send it. If "10 adults, 45 minutes" is retained, and the date and venue are still undetermined, that is an initial success. If you have advanced to the work or school entry point, do not try using work data as-is; verify your account. If you cannot find the new chat, use the updated chat input field; there is no need to recreate or search for undisplayed menus.

Why this topic matters

Act as an advisor to host a fictitious neighborhood book club. The goal is to create an announcement to send to participants, look up necessary items to bring, and set up a state where you can continue the same work later. Do not use real attendee names or addresses; test using only a practice-oriented virtual project.

The important thing is not just whether the AI gave a plausible answer, butbeing able to distinguish which information was input, what operations were actually performed, and what humans approved.Generative AI helps with drafting and creating options, but it is not a system that automatically guarantees fact-checking or final business decisions. Articulating success criteria beforehand makes it clear how many times you should try on the same screen and where you should stop.

Figure 1: Decision flow for beginners to avoid confusion

flowchart TB
 A["やりたいことを言葉にする"] --> B{"資料や画面が必要か"}
 B -->|"不要"| C["新しいチャットで質問"]
 B -->|"必要"| D["添付と権限を確認"]
 C --> E["回答と根拠を確認"]
 D --> E
 E --> F{"続けて修正するか"}
 F -->|"はい"| G["同じチャットで条件追加"]
 F -->|"別の話題"| H["新しいチャット"]

How to read the figure: This figure does not illustrate internal Microsoft implementations, but rather serves as a conceptual diagram to help users choose their actions. The diagram does not grant permissions or buttons that are not present in the actual interface. Product entry points and agreements can be checked inMicrosoft Copilot app update informationandLicense verification methods.

1. Copilot is not a single product

Even if items display the same blue or rainbow icon and are labeled "Copilot," the information available for conversation varies depending on what the user is signed into. The consumer version starts with your own inquiries or public information. The work and school versions operate under Entra ID and organizational management, where the scope of accessible work data changes based on assigned permissions and agreements. Beginners should first understand "which Copilot they are currently using" rather than the intended use case.

Order of testing on the screen

Open the profile icon to verify whether you are using a personal Microsoft account or a work/school account. Do not paste account email addresses into articles or external chats. Check contract details via your profile or the official account management page, and do not treat contract information guessed by Copilot itself as correct.

Prompts you can copy and paste

私の画面にあるCopilotの機能を知りたいです。個人向けCopilotと仕事向けCopilot Chatの違いを、アクセスできる情報・契約・安全上の注意で比較してください。私がログインしている実際のアカウントは推測しないでください。

How to determine if a result is correct

Check whether you are confusing the product category based on the profile display with general AI explanations.

Comparison by changing a single condition

Before testing a hypothetical question written for personal use with the exact same string on a work account, verify the contract and permission information for each screen. A difference in responses does not necessarily indicate a malfunction.

Failure and safety precautions

Do not jump to the conclusion that pasting workplace documents into the consumer Copilot will allow it to access work data. Prioritize your organization's data classification and cloud usage policies.

Primary Source / 1. Copilot is not a single product：Microsoft Support: Official explanation of this feature

2. Read the left menu based on "what you want to keep"

The "New Chat" on the left may be presented as the entry point to start a conversation, and "Library" as a place to find created items and referenceable content. "Discover" can be an entry point for finding usage examples, and "Imagine" may handle image creation. Screen names are not permanently fixed and can be changed or consolidated through updates. Distinguishing between the purpose and the storage destination before clicking everything helps prevent erroneous operations.

Order of testing on screen

Look at the actual left menu from top to bottom and write down only the names of the displayed items. At that time, add dates to items that change with updates, such as "Tasks (Preview)" or "Projects". Separate the fact that "something is displayed" from "being able to use the operation under the license".

Copy-and-pasteable prompts

初心者が迷わないよう、私の画面に『新しいチャット』『ライブラリ』『発見する』『Imagine』『タスク』『プロジェクト』があると仮定して、相談する・コンテンツを探す・画像を作る・後から続けるの4目的に分類してください。

How to determine if the result is correct

The distinction between consultation-based entry points and save/generation-based entry points is visible, preventing prompts that execute unintended sharing or purchasing from getting mixed in.

Comparison by changing a single condition

Change the objective from "thinking about a guide text" to "finding it later" and have it explain how the selection screen changes.

Failure and safety precautions

As noted in the Microsoft migration troubleshooting guideHandling Microsoft migration issuesolder screen headings should not be assumed to be current operations for everyone.

Primary source / 2. Read the left menu based on "what you want to keep"：Microsoft Support: Official explanation of this feature

3. Distinguish between new chats and continuing conversations

Asking follow-up questions within the same chat causes the prior context of that conversation to influence the response. On the other hand, starting a new chat when initiating a different topic reduces the risk of mixing in previous assumptions. Mixing work contract details into a conversation about travel packing makes it easy to lose track of the premises under which the AI is speaking. Conversations should be structured by task units.

Order of testing on screen

Open a new chat and enter "Create a 100-character notice for a fictional book club." Next, add "Rewrite it in terms understandable to an elementary school student" to the same chat and observe how the content is carried over. Then, open a new chat to request a separate shopping list and observe the difference.

Copy-and-pasteable prompts

架空の読書会の案内文を作ってください。対象は大人10人、所要45分、日時と会場は未定です。日時と会場を勝手に決めず、仮の記載だと分かるようにしてください。

How to verify if the results are correct

Check whether the 45-minute duration and headcount remain, and if unconfirmed items have been filled in as facts.

Comparison by changing a single condition

Check if the 45-minute duration still applies when changing the prompt to "for elementary school students" via an additional query. Do not assume that starting a new chat will automatically carry over the same premises.

Failures and safety precautions

Chat history and memory are separate features. Starting a new chat does not necessarily invalidate all past personal settings.

Primary Source: 3. Choosing between a new chat and a continuing conversation：Microsoft Support: Official explanation of this feature

4. What kind of mode is Smart?

"Smart," displayed in the input field, relates to the conversation mode of consumer Copilot. Official mode explanations distinguish between options for quick responses and options that take time to think. Microsoft's official help describes Smart as of October 27, 2026, as "a conversation mode powered by GPT-5." However, displayed modes and internal specifications are subject to update. Selecting Smart does not guarantee the accuracy of the response or up-to-date grounding. Check the official link in this chapter for detailed specifications.

Step-by-step testing on the screen

If "Smart" is displayed near the input field, open the selection menu before sending and check for other available modes. If not displayed, do not apply outdated explanations. Test the same hypothetical question in two available modes and compare differences in phrasing, follow-up questions, and sources.

Copy-and-paste prompt ready for use

雲ができる理由を、小学生にも分かる日本語で、①蒸発②上昇③冷却の3段階に分けて説明してください。専門用語には短い説明を添えてください。

How to verify if the results are correct

Check whether the specified three stages are answered and whether content requiring source verification is mixed in.

Comparison by changing a single condition

Evaluate the differences between the Smart response and the response from another mode based on missing key points and grounding, rather than length.

Failures and safety precautions

Read the explanation of Smart as the official Microsoft specification as of that verification date, and do not infer the reader's internal model or subscription status based solely on the display name. Available menus vary by time and subscription.

Primary Source: 4. What kind of mode is Smart?：Microsoft Support: Official explanation of this feature

5. Discovering and safely testing recommended use cases

Suggestions listed under "Discover" are usage examples tailored to the user currently viewing the screen. If you are a beginner, instead of drafting long prompts from scratch, you can look at these examples to learn how to input prompts. However, if a recommended use case involves sharing personal information or publishing data externally, you must not execute it as-is, but rather replace the details with fictional subjects. The suggested items have not yet been executed.

Step-by-step procedure for testing on the screen

Select an example displayed in the Discover section and change only the target of the prompt to a "fictional book club." Check the conditions of the response and verify whether real company names or customer names are being auto-completed. If no examples are provided, start directly from the chat input field.

Copy-and-paste ready prompt

架空の読書会で参加者が話しやすくなる質問を5個提案してください。参加者の実名を想像せず、誰でも参加できる一般的な質問にしてください。

How to determine if the result is correct

Check whether the specified quantity of five items, the difficulty level, and the scope of the target are maintained.

Comparison by changing a single condition

Observe how the questions and vocabulary change when changing the fictional target from "adults" to "elementary school students."

Failure and safety precautions

Even if usage suggestions include phrasing like "send this now," always verify whether approval for sending permissions is required.

Primary Source: 5. Discovering and safely testing recommended use cases：Microsoft Support: Official explanation of this feature

6. Realistic sequence for checking subscription agreements

There is not necessarily a single button to easily determine whether a plan is free or paid. Personal Microsoft 365 plans include categories such as Personal, Family, Premium, and Pro, and usage limits for voice, images, Pages, and other features vary by plan. For enterprise and work accounts, features may change depending on the applicable Microsoft 365 subscription, Copilot add-on license, administrator settings, or usage-based billing. It is important to check "what can be done" using feature-by-feature comparison tables.

Step-by-step procedure for testing on the screen

Verify your account by referring to the official documentation for "Types of Copilot licenses you hold" and "AI credits and limits." Select one feature you want to use and note down four items: the compatible license, region, administrator settings, and usage quota. If there is a discrepancy between your account subscription display and the product screen, verify it before making a payment.

Copy-and-paste ready prompt

個人用Microsoft 365 Personal、Family、Premiumと、職場向けCopilot Chatの違いを機能別に整理してください。料金や利用上限は公式の現行情報で確認できない限り固定値にしないでください。

How to determine if the result is correct

Check if the table separates available, conditional, and unverified items, and does not simply state that 'everything is possible with a paid plan'.

Comparison by changing a single condition

Compare how limits differ for separate features like image generation and Pages, even under the same license.

Precautions for failures and safety

Do not confuse shareable subscription benefits with AI benefits restricted to the subscription owner.

Primary Source: 6. Realistic Order for Verifying Contracts：Microsoft Support: Official explanation of this feature

7. First-Day Practice: Separate 'Proposals and Results'

Simply speaking in Japanese in the Copilot input field is not difficult. However, what matters in practice is distinguishing between what was requested and what has actually been completed. Even if the generated draft states that an 'email has been sent,' that alone does not mean the email has actually been sent. Verify whether Copilot followed the instructions by checking the body conditions, the actual status in related applications, and the user's operation history separately.

Order of testing on the screen

Create a fictitious announcement and mark fields with insufficient conditions as 'TBD'. Next, ask for '5 verification points before sending,' and then check on the screen whether the generated output is merely a draft or has been saved to another application. Do not actually send externally during experiments.

Copy-and-paste prompt ready for use

この案内文を家族に送る前のチェックリストを5項目作ってください。まだメール送信・共有リンク作成・ファイル保存はしないでください。

How to determine if a result is correct

Check whether items have been established to inspect the body text, deadlines, recipients, and undecided matters.

Comparison by changing a single condition

Compare whether changing the date and time from 'TBD' to 'tentatively a Saturday in November' changes the announcement and verification items.

Precautions for failures and safety

Do not confuse AI proposals with human confirmation. Wait for administrator approval for important matters.

Primary Source: 7. First-Day Practice: Separate 'Proposals and Results'：Microsoft Support: Official explanation of this feature

8. Moving to practical applications with four-element prompts

The initial prompt can be brief, but when consistent output is required, group and provide the objective, background, information sources, and output format. By specifying not only what you want done but also where to retrieve facts from, you can reduce the risk of the model inventing its own conditions. Official Microsoft prompt guides also explain perspectives such as goal, context, details, and data. Even if terminology varies depending on the teaching method, the underlying concept remains the same.

Sequence for testing on screen

For the same hypothetical proposal, start by simply asking, 'Plan a reading group.' Next, retry with explicit constraints: 10 participants, 45 minutes, zero cost, source materials limited to the provided text, and the output formatted as a table. A human must verify whether the sum of the time allocations equals 45 minutes.

Copy-and-paste ready prompt

目的：初心者向け読書会の進行案。背景：架空の参加者10名、45分。情報源：このメッセージの条件だけ。形式：時間配分の合計が45分になる表と未決事項3点。

How to verify correct results

Check whether the total time is correct, pending items are addressed, and no facts were arbitrarily invented.

Comparison by changing a single condition

Change only the time limit from 45 minutes to 60 minutes and observe whether the time allocation and priorities change.

Failure and safety precautions

If participant information not present in the input is generated, correct it, and do not assume that a meeting room or calendar reservation has been made.

Primary source / 8. Moving to practical applications with four-element prompts：Microsoft Support: Official documentation for this feature

9. If the UI changes, suspect an update before assuming a malfunction

Copilot menus are updated in 2026, with notices regarding the renaming of Projects to Notebooks and the migration from Tasks to Cowork. Even if the user's screen differs from screenshots in online articles, that alone does not indicate an account issue. During Microsoft rollout updates, latency may also occur in display rendering and history synchronization across devices.

Sequence for testing on screen

Record the menu names and account type on your current screen, and check the date on the official update page. After verifying the mapping between older names (such as 'former Projects') and current names (such as 'current Notebooks'), locate the target content. Also verify whether you are logged in with a different account.

Copy-and-paste ready prompt

以前は『プロジェクト』がありました。いまは見当たりません。改称や移行の可能性、現在のアカウント、同期状況について、優先順に確認事項を整理してください。

How to verify correct results

Check whether non-existent menus or fictional settings are being referenced.

Comparison with one changed condition

By adding the condition that the menu with the same name remains on another device, should rollout or synchronization discrepancies be considered?

Precautions regarding failures and safety

Do not delete existing data arbitrarily while referencing the old UI. Do not recreate content before migration and synchronization are complete.

Primary Source: 9. If the screen changes, suspect an update before assuming a failure：Microsoft Support: Official explanation of this feature

Troubleshooting matrix for when you get lost on the screen

Even if the left menu or mode name differs from the article, there is no need to immediately assume a failure. First,The account currently signed in、The displayed feature name、The permissions actually availableCheck these three points separately. Since guided features differ between personal and work accounts, it is important not to force the same steps to apply.

For example, if you request a book club announcement and only the text is displayed, that isa draft text state. Saving it as a file, sharing it with a third party, and registering it in a calendar are separate operations. Initially, limiting yourself to "reading the text displayed on the screen, copying it yourself, and choosing the save destination" makes it easier to track what has actually been completed.

A quick reference table of features for work use

This is Important: Even when using the same 'consultation' process, prompting the AI to draft text as if making paper notes is distinct from creating, saving, or sharing items within the user's Microsoft account. Do not determine permissions and execution based solely on the AI's response text. The fundamental rule to teach users first is recognizing this boundary.

Designing 'Verification Accountability' When Deploying to Production

While practice for individuals may only require a few questions, sharing a workflow as a standard operating procedure requires four elements: the procedure manual, permissions, a verifier, and halting conditions. Simply specifying who selects the source data, what scope is passed to Copilot, who reviews the basis for the output, and who transmits data externally reduces the risk of over-reliance on AI responses. For example, if the same dummy data can be used to test processes even when personnel change, and results can be visually inspected, the onboarding explanation becomes repeatable.

Verification does not end with simply checking whether the tool was used successfully. Finally, confirm whether personal information was unintentionally entered, whether information from unfamiliar external sites was mistakenly treated as authoritative, and whether data from a different account was referenced despite having the same file name. In particular, because whether Copilot references the latest information varies by query, users must refer back to original official documents for matters related to laws, financial amounts, contracts, and security settings.

Verification Sheet for Beginners

Common misunderstandings and countermeasures

Q. That feature is not visible on the screen. Will adding a subscription guarantee its appearance?

A. This cannot be guaranteed. Check the differences between personal and work/school accounts, subscriptions, administrator disable settings, region, previews, phased rollouts, and name changes/integrations in order. There is no need to repeat procedures that assume a button not present on the screen.

Q. Can I trust Copilot when it says it has saved or sent something?

A. The response text alone is insufficient. Verify the destination file or the sent list using the relevant app and permissions. The exercises in this article do not perform actual transmissions or file modifications.

Q. Does the accuracy of the response differ between the free and paid versions?

A. While limits on models, usage counts, and features may vary, being on a paid plan does not mean you can skip verifying the grounds. Ultimately, humans must check whether the provided materials were read correctly and whether dates and figures are accurate.

Q. Are all reference links official Microsoft operation procedures?

A. Reference links serve as the basis for features and provision conditions. The fictional scenarios, comparison tables, prompts, and checklists in the article are editorial suggestions and do not replicate official Microsoft operation screens.

What was clarified in this chapter

What truly matters in this theme is not how many buttons you know, butIt is about being able to explain what your Copilot can do and what you have not verified yet.Test with small dummy data, check the difference from the expected output, and return to the source if there are any issues. Repeating this method will make it easier to avoid confusion when the UI or AI model changes. Do not generalize the procedures and diagrams described in this article to users' contracts, operating systems, or application versions without permission.

Solving "too many items in the left menu" with an end-to-end walkthrough on day one

If you try out the screen's function names sequentially from top to bottom, you are likely to lose track of what you were trying to accomplish in the end. When using Copilot for work, it is better to decide "what to create" first and only navigate to the necessary areas. Here, we will prepare a single fictitious reading group session. We will connect everything from consultation, revision, history check, and destination determination as a single small experiment.

Practice 1: Round 1: Intentionally asking vaguely to see what is missing

Type "Create an event announcement" in a new chat. Because you have not specified the capacity, date and time, or target audience, the AI may make arbitrary assumptions or return questions. Do not consider this a malfunction; instead, extract the premises included in the result. For example, if it says "next Saturday" or "free admission," those strings do not exist in the input. This is an example of the ambiguity of your request transferring to the output.

What to verify locally: Organize the missing conditions and re-enter. Request that it be based solely on the facts: "Capacity of 10 people, 45 minutes, venue, date and time, and fee undecided." Check whether the date has been arbitrarily finalized even in the improved text.

Practice 2: Round 2: Changing conditions in the same conversation

Change the capacity from 10 people to 12 people. When you ask Copilot to "revise the previous text," check whether the capacity in the new draft is 12 people. At this time, it is natural for the previous response to remain as 10 people. The conversation history represents the history of changes and is not necessarily a mechanism that always displays only the latest official version.

What to verify locally: Clearly define the target you are copying as the "last verified version." If both texts remain, briefly ask again to "list the confirmed items of the latest version" and select the verified items.

Practice 3: Round 3: Starting with Smart and changing only when necessary

If Smart is displayed near the input field on the screen, start by asking questions in that mode. If you can switch to another mode, use the same fictitious question and observe what differences occur in the response format, depth, and facts that require verification. Switching modes does not automatically eliminate the need to check sources. Separate consultations that require rapid responses from those where you want to spend time on complex comparisons.

What to verify locally: When grading two responses, do not automatically assume "short equals low quality," but check whether important conditions are met and whether comparisons and grounds can be verified.

Practice 4: Final: Choosing where to leave it

Once the final draft of the announcement is ready, check whether the current chat can be revisited from the history. If you have a contract where Pages is available, you have the option to move to an editable document. On the other hand, you can also achieve your goal by simply copying the text and leaving it in a local memo. Even if an item called Library is visible, it does not necessarily mean the announcement was automatically saved as a file.

What to verify locally: Categorize the necessary deliverables into "chat history," "official announcement," and "referenced conditions," and verify the save destination. Do not publish or send yet.

Do not infer the operation of the new UI from the display of the old UI

Descriptions of the old UI include terms such as New Chat, Library, Tasks (Preview), Projects, Discover, Imagine, Experiments, and the + and Smart buttons in the input field. However, user screens were not observed during this revision. Even if text is displayed, user subscriptions, internal mode selections, remaining limits, and experimental feature details cannot be definitively determined.

Because this is a long article, we want to avoid driving readers to search for non-existent menus. Treat facts observed in screenshots, specifications officially stated as provided, and operations successful in the reader's own environment as distinct pieces of information.

Evidence to retain in this practice

• Current Copilot type (personal/work) and verification date/time

• The actual menu name displayed. Distinguish it from the text in previous introduction images

• The fictitious conditions and prompts entered, along with the main part of the AI's response

• Discrepancies from expected values, re-entered conditions, and remaining unverified items

• Record sending, file updates, and sharing only when actually verified, anddo not mark untried processes as completed

The pitfall here is assuming that the moment you open Copilot, you can ask it to do anything. Available features vary depending on the product, subscription, and permitted data. Conversely, there is no need to assume from the start that an expensive plan is required. Start with the practice of organizing fictitious text, and verify the conditions with Microsoft's official documentation only when there is a specific feature you want to use.

Official information and primary sources (verified October 27, 2026)

• 1. Microsoft Support: Official documentation for updates

• 2. Microsoft Support: Official documentation for licenses

• 3. Microsoft Support: Official documentation for limits

• 4. Microsoft Support: Official documentation for getting started

• 5. Microsoft Support: Official documentation for modes

• 6. Microsoft Support: Official documentation for prompts

• 7. Microsoft Support: Post-update content and Projects migration

Verification limitations: Although the specifications and public descriptions in primary sources were cross-checked against the manuscript, the prompts featured in the article were not entered into an actual Microsoft Copilot account. Verification of physical device screens, target licenses, currentness of screenshots, WordPress rendering, independent AI review, and duplication checks including unpublished posts has not been completed.