Separate chat, Build, Bot and Imagine before spending time or credits.
What you’ll finish with A small first task in the Grok tool that fits it.
AI-generated guide · Content revised · Allow about 15 minutes for the exercise
The walkthrough
Start with the output
Use chat for an answer or analysis, Build for a coding task, Bot for work that needs a persistent computer, and Imagine for image or video generation. Check current access in your account.
Make success observable
For research, require supporting sources. For code, name a test and inspect the diff. For media, specify dimensions and check the final export. Avoid a vague instruction to make something better.
Set a small first scope
Start with a short document, one bug or one short clip. Connect only the accounts needed for the task. Keep external actions such as sending messages or publishing subject to your approval.
Check the total allowance
Inspect the usage meter after the trial. Since October 12, Grok and Cursor draw from one shared pool on The X Subscription, so coding work uses the same allowance as chat; API requests are billed separately. Measure the work left for you before paying for more volume.
Text to start with
Replace the bracketed sections with your own details.
Task: [specific output]. Sources or files: [material]. Success checks: [checks]. Work only within this scope and ask before any external action. At the end, list what you completed, what you verified and anything unresolved.
A complete worked example
Choose the output first. Feature availability, allowances and connector permissions can vary by account; confirm them in the current product documentation before committing to a workflow.
Choose a surface
Swipe or scroll across the table to see every column.
Choose a surface
Surface
Concrete first task
Check before granting access
Chat
Summarize a public announcement with links and dates.
Open the cited sources and check support for each claim.
Build
Fix the zero-quantity bug in the coding guide.
Limit repository scope; inspect the diff and run the regression check.
Bot
Prepare a comparison from three public webpages in its computer workspace.
Connect only the accounts needed; approve external actions such as sending or publishing.
Imagine
Create one short shot from the image/video guide.
Confirm current duration, export options, reference rights and credit usage.
A complete first brief
Compare these three public pages about the same announcement.
Make a table of confirmed changes, publication dates and disagreements.
Link each claim to its supporting passage.
Do not sign in, connect accounts, send messages, buy anything or publish.
Leave unsupported details unresolved.
Finish with the checks you actually completed.
Know when to change tools
Use chat when an answer from supplied material is enough. A coding task needs code access and a test, which favors a coding surface. Work in a persistent computer needs a clear stop condition and an action boundary. Media generation needs a creative brief and export checks. Do not add connected accounts merely because the interface offers them. Record whether the trial used an app allowance or separately billed API work.
When it goes wrong
A broad request turns into account connections or external actions you did not intend.
How to correct it
Restate the output, permitted sources and stop condition. Narrow the task and remove unnecessary permissions before another run.
Finished when
The chosen surface produces the requested artifact, stays within scope and reports unresolved work honestly.
Check the result
Did the tool actually inspect the supplied material?
Are sources and generated facts verifiable?
Which subscription or API balance paid for the task?
Before paying
Start with the access already in your account: Build is on every plan, Grok Bot starts at the $30 Premium tier and 1080p video at the $100 Super tier. Check the shared allowance and separate API billing before upgrading.