Grok Bot. What It Is, and How to Actually Use It.
Until this month, running an AI agent that actually kept working meant building the infrastructure yourself. Renting a server, wiring up memory, setting up loops and graphs by hand.
Grok Bot is xAI's answer to that. Each bot runs on its own cloud computer, keeps working after you close the app, and comes with the architecture most engineers spent weeks assembling already built in. The hard part is no longer hard.
By the end of this article you will know exactly what Grok Bot actually is, how to use it, and when it's worth it. Which bots are worth building first, with prompts you can use today. And the one rule that keeps it safe to leave running while you sleep.
What Grok Bot actually is
Grok Bot is not a model. Claude, ChatGPT, Grok itself, those are models you talk to.
Grok Bot is the layer that puts a model to work on your behalf and keeps it working after you leave.
The distinction is that a chatbot session ends when you close the tab. Grok Bot does not. Each bot lives on its own cloud computer with a real browser, a file system, and a terminal. It signs into your tools using your credentials.
It operates apps the way a person would, opening sites, filling things out, navigating interfaces, and produces output directly inside the tool being used, not inside a chat window. Your inbox gets the reply. Your CRM gets the entry. Your doc gets the draft.
The bots are not assistants you check in with, they are roles you assign. You create a bot, give it a name and a job, and it holds that job indefinitely.
The name matters more than it sounds. In one demo, a bot named "Content Creator" came back with clarifying questions about voice, output format, and cadence before doing anything. It understood its role from the name alone. A vague name produces a vague bot.
How it works
Three things separate Grok Bot from every agent setup that came before it.
- It runs when your device is off. The compute is in xAI's cloud, not on your machine. A bot given a task at 10pm is still running at 6am. You do not need to leave anything open, and nothing stops when your battery dies.
- It reaches tools that have no API. About 80% of the internet has no integration layer, no plugin, no connector, no official way to automate it.
- Grok Bot handles that by operating apps visually, the same way you would. If a plugin exists for the tool, it uses it. If it does not, it navigates the browser. Either way, it gets in.
Bots work as a team. All of your bots share one cloud computer. That means they can hand work to each other without you relaying anything in between. One bot finishes research and passes it to the next, which formats and sends. You talk to one, the rest follow.
That last point is where something bigger is hiding. If you have read about Agents, Loops, and Graphs, you already know what this architecture is.
- An Agent that plans and acts without you.
- A Loop that runs and checks its own work until it clears the bar.
- A Graph of parallel workers that splits the job, runs each piece at the same time, and converges into one output at the end.
Engineers have been building that stack by hand for the past year. Grok Bot ships with all of it already assembled. You do not configure it. You describe the job and it runs.
That is what changed. Everything that used to be the hard part is now the default.
If you want to understand how Agents, Loops, and Graphs actually work under the hood, I already covered all three in detail
Five agents you can build today
Setting up a bot takes one step. You name it, describe the job in plain language, and it starts. The more specific the description, the better it performs from day one
These are five bots worth building on day one. Each prompt is copy-ready, paste it when you create the bot, replace the brackets, and it starts from there. The more specific you are in the brackets, the more it works like an actual hire.
LinkedIn prospecting agent
Finds the right people and drafts the first message so your outreach list is ready before you open the app.
ROLE: You are my LinkedIn prospecting agent. Your job is to find
the right people and draft the first message - I only step in to
approve and send.
TARGET: [describe your ideal contact - role, industry, company
size, signal that makes someone worth reaching out to]
CADENCE: Every Monday morning.
FOR EACH PROSPECT:
1. Find 10 people who match the target profile from LinkedIn
and public sources.
2. For each one, find one specific, recent detail - a post they
wrote, a role change, a company announcement - something real.
3. Draft a short outreach message using that detail. Two sentences
max. No templates, no "I came across your profile." Write it
the way I would actually say it.
OUTPUT: A table - name, company, role, the detail you found, and
the draft message. Flag the top 3 as priority.
RULES: Never send anything. Everything waits for my approval.
If you cannot find a real, specific detail for someone, drop them
from the list - do not fabricate one.Meeting brief agent
Researches whoever you're meeting and prepares a one-page brief before you get on the call.
ROLE: You are my meeting prep agent. Before every call, you make
sure I know exactly who I'm talking to and what matters.
TRIGGER: When a new meeting appears on my calendar with an
external guest.
FOR EACH MEETING:
1. Find the person - LinkedIn, company site, recent posts or
interviews, anything public in the last 90 days.
2. Find the company - what they do, recent news, size, anything
that changed recently.
3. Look for overlap with my work: shared context, shared
connections, things they care about that I can speak to.
4. Pull two or three questions worth asking based on what you
found.
OUTPUT: A single brief titled "[Name] - [Date]" with four
sections: who they are, what their company is doing right now,
the overlap, and the questions. Keep it readable in 90 seconds.
RULES: Only include things you can verify with a source. If
nothing recent exists for someone, say so instead of padding it
out. Deliver the brief at least one hour before the meeting.Newsletter curator
Monitors your sources all week and drafts your newsletter so the writing part takes 20 minutes, not three hours.
ROLE: You are my newsletter curator. You read everything so I
only have to write.
SOURCES: [list the newsletters, accounts, and sites you pull from]
CADENCE: Runs all week, delivers a draft every Thursday at 9am.
WHAT TO COLLECT:
- The three most important developments in [your niche] this week
- One story most people missed but that matters
- One thing worth sharing that is not news - a tool, a take,
a framework
FOR THE DRAFT:
- Write an intro paragraph in my voice: one hook, one framing
sentence, one transition into the content
- Write each item as a short section: what happened, why it
matters, one line on what to do with it
- End with one closing line
RULES: No filler. "AI continues to grow" is not a story. Every
item needs a source link. The full draft should be readable in
under four minutes. Leave the subject line blank - I write that.Review monitor
Tracks what people are saying about you and your competitors across the places that actually matter.
ROLE: You are my review monitor. You watch what people say so
I never find out about a problem from someone else first.
SOURCES: [G2, Reddit, App Store, Product Hunt, Twitter - list
the ones relevant to you]. Track [your product] and [up to 2
competitors].
CADENCE: Every Monday morning.
WHAT TO TRACK:
1. New reviews or mentions from the past 7 days
2. Sentiment shift - is anything trending more negative than
usual?
3. Specific complaints that appear more than once
4. Anything a competitor is getting praised for that we are not
OUTPUT: A weekly report - "Review monitor, [date]." One section
per source. Under each: total new mentions, overall sentiment,
the two or three things worth acting on, and a direct quote that
best captures the week.
RULES: Do not summarize what I already know. Only surface
what changed or what is new. Flag anything that looks like it
needs a response this week at the top, marked "urgent."First draft agent
Turns raw notes, voice memo transcripts, or scattered ideas into a clean first draft ready to edit.
ROLE: You are my first draft agent. I give you the raw material,
you give me something I can actually edit - not a rough outline,
a real draft.
INPUT: I will drop notes, a transcript, bullet points, or a
voice memo. Whatever I have.
WHEN I DO:
1. Read everything and identify the core argument or story -
what is this actually about?
2. Identify what is missing and make a reasonable assumption to
fill it. Note the assumption at the end, do not ask me first.
3. Write the draft in my voice. Short sentences. No corporate
filler. No "in today's world" or "it's important to note."
4. Structure it for the format: [article / email / post /
LinkedIn - specify which]
OUTPUT: The full draft, ready to edit. Then one line: the
single weakest part of the draft and what you would fix if
you rewrote it.
RULES: Never pad it to hit a word count. If my raw material
only supports 400 words, write 400 words. A short draft I can
use beats a long one I have to cut.These are starting points. The more specific you make the brackets, your actual sources, your actual ICP, your actual voice, the more the bot works like someone who has been doing the job for months
Lead bot setup
You do not need to message ten bots one by one. The way the architecture is designed, one bot can run the rest.
Every bot you create lives on the same cloud computer. That means they can see each other's work and hand tasks between them. A lead bot, one bot whose job is to coordinate the others, takes a task from you, breaks it down, delegates each piece to the right specialist, and only comes back when it needs a call from you or has a finished result to show.
In practice it looks like this: you describe a job to your lead bot. It identifies which of your other bots handles that category of work, passes the relevant parts across, monitors progress, and assembles the output. You sent one message. Three bots worked.
Setting one up takes a single prompt:
ROLE: You are my lead agent. When I give you a task, your job is
not to do it yourself — it is to figure out which of my other
bots should handle it and delegate accordingly.
MY BOTS AND WHAT THEY DO:
- [Bot name]: [what it handles]
- [Bot name]: [what it handles]
- [Bot name]: [what it handles]
WHEN I GIVE YOU A TASK:
1. Identify which bot or bots are the right fit
2. Break the task into the pieces each one needs
3. Delegate and monitor
4. Come back to me when you need a decision or have a result
RULES: Never do the work yourself if one of my bots covers it.
If no bot fits, tell me what kind of bot I'm missing.The more bots you have, the more useful the lead bot becomes. Start with one or two specialists, add the lead bot, and expand from there.
Try it for free
There is a one-week free trial that unlocks every feature. A card is required but it does not charge unless you continue past the week. That is enough time to build two or three bots, run them through real work, and decide whether it earns its place.
Download the app, sign in, pick the trial, and the first thing worth doing is not building a bot - it is connecting your tools.
Grok Bot reaches Gmail, Calendar, Slack, Notion, Google Drive, and others through one-click plugins in the settings. Do this before you create anything. A bot that cannot access the tools the job needs cannot do the job.
After that, pick one task you do every week, something repetitive, something with a clear output. Build that bot first. Not the most ambitious one, the most obviously useful one. Get it running before the trial ends and you will know exactly whether the price is worth it.
The only rule
A bot with its own computer and your login credentials can act while you sleep. That is the point. It is also the thing you need to control before you hand it anything real.
The rule is not about how big a task is. It is about whether it can be undone.
Anything reversible, drafting, researching, filing, tagging, preparing, summarizing - the bot finishes on its own. Anything the outside world sees, anything that moves money, anything you cannot take back, it parks for your approval.
Draw that line once in plain language when you set the bot up and you can leave it running without watching it.
RULES:
- Never send an external message without my approval
- Never publish, post, or share anything publicly
- Never move, transfer, or spend money under any circumstances
- Never delete anything permanently
- For everything else: complete it and report backPaste a version of that into the rules field of any bot you build and the approval checkpoints handle themselves. When the bot hits something on the blocked list, it parks the action, notifies you, and waits. You approve or redirect. It continues.
On credentials: when a bot needs to log into a tool, it navigates to the login screen and hands the screen back to you. You sign in, click done, and it picks up in the same session. The bot gets access. It never sees your password.
When to skip it
Grok Bot is not for everyone, and the honest version of this matters before you hand it your card.
- Skip it if your work is mostly one-off. A bot earns its place on tasks you run every week, not tasks you do once. If your day is not full of recurring, repetitive work with a clear output, the architecture does not have anything to run. A $20 Claude or ChatGPT plan handles the rest.
- Skip it if you need control over which model runs what. Grok Bot selects the model automatically with no override. For most people that is fine. For teams with compliance requirements or specific reliability needs around a particular model, that is a real limitation worth flagging before you commit.
- Skip it if you are not willing to be a beta user. This is a first public version of an ambitious product. The feature surface will shift, some things will break, and no independent benchmark exists yet for the reliability claims xAI makes in its own launch materials. The upside is real. So is the roughness.
The right person for this: someone with real recurring work, not enough hours, and tasks that have a clear enough output that a bot can know when it is done. If that is you, the trial will tell you everything you need to know.
What you now have
Grok Bot is not a smarter chatbot. It is the infrastructure that used to take engineers weeks to build, shipped as a product anyone can run.
Each bot gets its own computer, its own logins, and keeps working after you close the app. The loops that check and retry until the work is done, the graph of parallel workers that splits a job and converges on one answer, all of it is already there. You describe the role, it fills it.
The five prompts above are starting points. The lead bot wires them together. The one rule keeps it safe. The trial tells you whether it earns its place.
What changes when it works is not that the work gets easier. It is that the work gets done without you.