Run a team of AI agents in Grok from your phone and computer in 7 minutes
Three roles - researcher, analyst, editor - work in one Grok chat and deliver a finished result for your task
- 1.Open Grok in your computer browser (grok.com) or in the X mobile app (the «Grok» tab on the bottom bar). Sign in with an X account or a new account - registration is free.
If the service is unavailable in your region, use a VPN. Grok's free tier is available to all registered users.
Grok ↗About this tool - 2.In the message input box, make sure the unrestricted mode is selected - the mode toggle should not be in the Think position. Just start typing the prompt below into the regular chat field.
The free tier lets you send about 20 - 25 messages every 2 hours - enough for one full cycle of a three-agent team.
Grok ↗About this tool - 3.Copy and send the following prompt into the chat - it sets up a team of three agents and starts them working on your topic. Replace [TOPIC] with your task.
One prompt launches all three roles at once - no need to create three separate chats.
Grok ↗About this toolPromptAct as a team of 3 AI agents working in sequence on this topic: [TOPIC] Agent 1 - Researcher: find 5 key facts, data points, or examples about the topic. Output as a numbered list. Agent 2 - Analyst: take Agent 1's output, identify the 3 most important insights, explain why they matter, and flag any risks or contradictions. Agent 3 - Editor: take Agent 2's output and write a final summary in plain language, 150 words max, with a clear action recommendation at the end. Run all three agents in this one response. Label each agent's section clearly: [RESEARCHER], [ANALYST], [EDITOR].
What this prompt doesThe prompt is in English. It makes Grok play three roles in a row in one response: the researcher gathers facts, the analyst highlights the key points, the editor writes the final summary. Replace [TOPIC] with your question - for example, «trends in remote work tools 2026» or «how to choose a CRM for a small business». - 4.Wait for the full response - it will be long and split into three sections labeled [RESEARCHER], [ANALYST], [EDITOR]. Scroll to the end and read the final [EDITOR] section - this is the finished result.
If the response cuts off due to a length limit, type «Continue from [EDITOR]» - Grok will pick up where it stopped.
Grok ↗About this tool - 5.To refine the result, send a follow-up prompt in the same chat - it addresses a specific agent. Replace [QUESTION] with what you want to clarify.
The context from the previous response is saved - the agents «remember» their work. You can address a specific agent by name.
Grok ↗About this toolPromptAgent 2 (Analyst), please go deeper on this point: [QUESTION]. Then Agent 3 (Editor), rewrite the final summary with this new information included.
What this prompt doesThe prompt is in English. It asks the «analyst» to dig deeper into a specific point, and the «editor» to rewrite the final summary with the new data. Replace [QUESTION] with what you want to clarify, for example: «compare costs of Notion vs Monday.com». - 6.Copy the final result: on a computer - select the text of the [EDITOR] section and press Ctrl+C / Cmd+C; on a phone - long-press the text, then «Copy». Paste it into your notes, a document, or a messenger.
To repeat the cycle for a new topic, start a new chat using the «New conversation» button or the «+» icon - this will reset the previous team's context.
Grok ↗About this tool
- freeGrok (free tier)The free tier is available to all registered X users; the limit is about 20 - 25 messages every 2 hours, which is enough for one full cycle of a three-agent team.
- subscriptionGrok SuperGrokThe paid subscription increases message limits and gives access to a more powerful model; current prices are on grok.com.
AI services change fast - interfaces and free limits may differ from what's described.