Guide Library

Week 1 · Guide 6 of 7

Foundations

Choose the Right Effort for Your Task

Field Guide

  • 15–20 minGuided walkthrough

Your agent may offer a quicker answer or more time to work through a difficult request. Try one small task, check the result, and decide whether asking for more effort would help. You can keep the normal setting when it does the job.

You’ll leave with

A checked result for your task and a short rule for when you'll use normal effort or ask for more.

What You Need for This Guide

  • Your assigned ChatGPT, Claude, Microsoft 365 Copilot Chat work account, or Gemini app with your Google Workspace account.
  • A small task and the facts or notes it needs. A computer helps when comparing two chats.
  • Your or the Project/Notebook you saved it in, for the task finder in step 1.

Create your Agent Profile

Choose the tool you’re allowed to use

Your approved tool

The tool your company approved.

Pick your approved tool above to see the steps and screenshots for it.

Start guide
5 steps · At your own pace
About saved progress

No account needed. Checkmarks save on this browser and device when browser storage is available. If storage is blocked or full, new checkmarks last until you reload or close this page. Progress doesn’t sync to other browsers or devices. Clearing this site’s browser data removes it, and private browsing may not keep it after you close the window.

Standard Data rules for AI at work

Data Rules

Data rules for AI at work

Check with your IT Team before you paste any work data into an AI tool. Avoid names, account numbers, customer data, or anything sensitive without explicit permission from IT. If you aren't sure, redact anything sensitive and replace it with generic information. AI makes mistakes, make sure to fact check your agents outputs even if they sound right.

Before you start

  • Check with your IT teamAsk what your company allows before you put anything from work into an AI tool. Your company policy is the one that counts, and it is the one this library cannot see.
  • Use the tool your company approvedChatGPT, Claude, Microsoft 365 Copilot Chat and Gemini Enterprise are all covered here. Use the one your company approved, not the one you like best.
  • Sign in with your work accountA work account keeps your chats inside the agreement your company signed. A personal account in the same product sits outside it.
  • No approved account yetFollow the steps with made-up example text, or on paper, until IT gives you one.

Never put this into an AI tool

  • Customer or personal detailsNo names, email addresses, phone numbers, home addresses or account numbers. That covers colleagues as well as customers.
  • Regulated recordsHealth, financial, HR and legal records stay in the systems built to hold them, unless IT has told you in writing that this tool may see them.
  • Passwords, keys and access codesNo credentials of any kind, yours or anyone else’s, and nothing that grants access to a system.
  • Anything under an NDA or not yet publicUnreleased plans, contracts, pricing and source code need clearance before they go near a chat window.
  • Anything your company policy prohibitsYour company’s rules win over anything in this library. If policy keeps something out of AI tools, it stays out, whatever a guide suggests.

Do this instead

  • Swap real details for placeholdersWrite “Customer A”, “[client name]”, “$X”. You get the same quality of answer, and you put the real details back yourself.
  • Describe the situation rather than pasting the documentMost of what an agent needs is the shape of the problem, not the file it arrived in.
  • Trim a file before you upload or paste itTake out the identifying columns and pages first, or work from a short extract.

Check the answer before you use it

  • Read it as if a colleague had drafted itAn agent states wrong things with complete confidence. Your name goes on the work, so the check is yours.
  • Say where the agent helpedFollow your company rule on disclosing AI assistance, especially in client-facing or formal work.

When you are not sure

  • Ask before you paste or upload, not afterA question to your IT team takes a minute. Cleaning up after a data incident takes a great deal longer.
  • When in doubt, leave it outIf you would not put it in an email to someone outside the company, keep it out of the chat.
Open the full page
Download Guide

Step

Choose a small task and two checks

Choose a task you can judge from information you have: rewriting a handover note, comparing two supplied options, or planning an afternoon at home. Aim for one short result. Leave tasks needing outside research for another session.

Write two checks in your notes. Name something the answer must preserve and something it must achieve, such as keeping every supplied date and fitting into five bullets. For a decision, name the conditions an acceptable option must meet.

Find a small task to try
Help me choose one small task for trying my agent's effort settings. It should produce something I could use and check from facts or notes I already have. Use my Agent Profile if I provide it, but do not require one.

Ask one short question at a time. Offer two or three concrete possibilities as numbered choices, plus “Something else / none of these,” and let me answer in my own words. Treat suggestions as possibilities, not tasks you assume I have. Ask what information I can supply. Avoid tasks needing outside research or actions in other apps.

Stop searching when I choose a task. If I already name one, use it. Help me settle on two checks I can apply to its result. Suggest checks suited to my task and available information, and wait for me to confirm or change them.

Return one code block with two headings: “What I need” for my chosen task, and “What done looks like” for my two confirmed checks. Use only confirmed details. Do not invent facts or requirements. Stop without doing the task.

Step

Find the controls for your tool

A model is the AI system answering you. An effort setting asks a supported model to spend more or less work on its answer. Some tools offer response modes that can also change the model. Their labels don't form one shared scale.

Open a new chat and note your current choice so you can restore it later. Start with the everyday or recommended option below when available. Follow only your tool's instructions.

In Chat, open Thinking effort beside the message box. The captured account has a Power slider with Medium selected. Start there if your menu matches. For a comparison, move one position higher and keep the separate model choice unchanged.

Thinking effort opens this account's Power slider. Medium is selected. The model choice is separate.

Open the model selector beside the message box, then Effort. Use Medium if offered. For a comparison, choose High with the same model. The captured Opus 5.5 menu also offers Low, Extra and Max. You don't need the maximum here.

Open the selected model to find Effort.
This Opus 5.5 menu offers Medium and High alongside other choices. Availability depends on the model and account.

In Microsoft 365 Copilot Chat with your work account, open the response selector above the chat. Start with Auto. For a comparison, choose Think deeper. Quick response is another everyday option. Auto can already choose deeper reasoning, so the comparison doesn't guarantee a different model.

The work-account menu offers Auto, Quick response and Think deeper. Auto can already route a request to deeper reasoning.

In the Gemini app with your Google Workspace account, open the model name in the message box. The captured choices are 3.6 Flash, 3.6 Thinking and 3.1 Pro. Start with Flash, then choose Thinking for a comparison if present. These are app choices, not a universal effort slider.

This work-account menu offers 3.6 Flash, 3.6 Thinking and 3.1 Pro. Flash is selected.

Menus vary by account and change over time. If these choices are unavailable, continue with your available setting and skip the comparison. Screenshots show menus observed on September 30, 2026.

Step

Try the task at the normal setting

Copy Brief your agent into the new chat from Step 2. Put your task under What I need and your two checks under What done looks like. If you used the finder, copy those two sections from its note. Add the background under What is going on and the reader under Who it is for. Include the facts or notes the task needs. Save a copy of the completed message in your notes, then send it.

Brief your agent
Download
If my Agent Profile is available, use it as background.

What is going on:
[Two or three sentences of background, the way you would tell a new colleague on their first day.]

What I need:
[The thing you want back: an email, a summary, a plan, three headline options.]

Who it is for:
[The reader, and what they care about.]

What done looks like:
[Length, format, and anything it must include or must not say.]

Use the facts I supplied. Do not invent dates, approvals, owners or commitments. Mark missing facts clearly.

Check the answer against your notes. If an important fact is missing, supply it in the same chat and add it to your saved message. If a detail was invented, ask for a revision using only supplied facts. After either correction, get an updated answer and recheck it before comparing settings. A question or marked unknown can be appropriate. If you can't supply a necessary fact, pause and gather it first.

Step

Decide whether more thinking is worth a try

Keep the normal setting when the answer meets your checks. Rewording notes or organizing supplied facts often needs no comparison. You can go to Step 5. For a wording or length miss, ask for that correction at the normal setting first.

Try more thinking when several conditions affect each other, a comparison is difficult, or a requirement is missed despite complete input. More work may help with reasoning, but may take longer or use more of your allowance. It cannot make unsupported facts true.

  1. Try the other setting once

    Open a second new chat in the same tool and choose Step 2's higher-thinking option. Paste your saved message with any clarification. Include the same relevant background, including content the task needs. Keep the first answer out of this chat.

  2. Compare against your checks

    Check both answers against the same two requirements. Keep the one that meets them with less work from you. If both work equally well, the normal setting is sensible. Longer wording or greater confidence doesn't win. After necessary clarification or correction, one comparison pair is enough.

If neither works, identify missing facts, unclear requirements or conflicting conditions before trying again. More effort may not solve that problem. Automatic routing and ordinary variation can change replies too, so this comparison supports a practical choice, not proof that effort caused the difference.

Step

Keep a rule you can use next time

Copy the checked result into the notes document holding your completed prompt. Add the task, the setting you will start with next time and when you would consider trying more thinking. If you kept the normal setting, note what worked. Restore your original setting if you changed it only for this exercise.

Use that rule for similar tasks. Recheck it when the task or available controls change. Reuse Brief your agent in a new chat with the new facts, and edit the prompt to suit your work.

Choose the Right Effort for Your Task

Confirm your result

Confirm the outcomes below to finish this guide.

Final checks

I have a result for my task that meets the two checks I chose.

I have a starting setting for similar tasks and a specific reason to try more effort.

Provide Feedback

Thanks. That went straight to the people who write these guides.

Next in Week 1

Get Unstuck with a Screenshot

Continue in the suggested order, or choose another guide for your task.

View Week 1 guides