The fan-out method: three models, one brief

How this helps you

You compare three models' attempts at the same job side by side and keep the best parts of each.

✦ Try it

Go ahead and use the cloud agent and launch three models: [model 1], [model 2], [model 3]. Same brief for each. Once you're done with everything, send me screenshots and screen recordings of each so I can compare.

What they did

From Joseph Yang in the official Grok Bot 201 workshop, via @GrokBotRadar: ask your Bot to launch Cursor cloud agents on three models with one brief. They build at once, you get screenshots and recordings of each, and you keep the best parts.

Who it's for: People building apps or prototypes who want to compare models.

Before you start

Cursor connected to your Bot.

How to try it

Connect Cursor to your Bot.
Write one clear brief for what you want built.
Ask the Bot to launch three Cursor cloud agents on different models with the same brief.
Ask for screenshots and screen recordings of each result.
Pick the best parts and ask one agent to combine them.

What to expect

Screenshots and screen recordings from each model's build. In the workshop example, a sticker shop, one model's drag-and-drop was the part kept.

Common mistakes

Using a vague brief. All three agents need the same clear target or you can't compare them.
Running it on small tasks. Three builds cost three times the usage.

Related tips

Watch coding work live and get recordings back: With Cursor connected, your Bot links you to the cloud agent live and sends screenshots and screen recordings.
How to trust your Bot's work more: Put most of your effort at the top: code structure, then lint/tests/CI, then prose rules, then human review.
Give deep research a finish line: "Research this deeply" gives no finish line and no reason to stop browsing.
A daily loop: a metrics Bot decides, other Bots change the site: In the official 201 workshop, an orchestrator Bot runs a daily skill: read the analytics, drop the bottom three, design three new, deploy.