Grok Bot examples: what people actually use it for
Grok Bot is xAI's agent product, in beta since August 11, 2026. Each bot gets a role, its own logins and a cloud computer, and keeps working after you close your laptop.
293 posts, videos and repos from X, LinkedIn, YouTube and GitHub, plus 30 prompts you can copy, sorted by use case. Numbers are each poster's own claims. We include the limits and the failures.
30 copyable Grok Bot prompts, from the people who used them
Each prompt is copied word for word from a public X or LinkedIn post, with the author and a link to the post. Placeholders in square brackets are the author's own. We have not tested these, so run them on a low-risk job first.
Email and inbox
Work-inbox reply assistant (start of a longer instruction)
You are my work-inbox reply assistant running as a Grok Bot / Automation.
Primary goal: Scan my work inbox for messages received since the last successful run of this job that plausibly need a human reply from me. Produce short digests + proposed replies in my voice. Never send anything. If nothing qualifies, produce zero output.
Written by mr fundman (@mrfundman) on X, Aug 16. Copied word for word from the post, first part of a longer prompt. AY Automate has not tested it. Source post
Email and inbox
Scan the inbox and draft replies
Scan my work inbox for messages since the last run that plausibly need a reply: customer and prospect threads, renewal or pricing questions, intros, direct questions to me. Skip newsletters, automated notifications, receipts, calendar RSVPs, and internal noise. For each one, give me a short digest in chat: sender, subject, one line on what they need, and a proposed reply in my voice. Do not auto-send. If nothing needs a reply, send no message at all.
Written by Kirill (@kirillk_web3) on X, Sep 3. Copied word for word from the post. AY Automate has not tested it. Source post
Outreach and sales
Follow-up drafts from call notes
Find external calls that ended since the last run, meaning any meeting with an attendee outside my company's domains. Meeting notes are the source of truth. Draft a follow-up in chat with To, Subject, and body, grounded in what was actually discussed, with concrete next steps.
Written by Kirill (@kirillk_web3) on X, Sep 3. Copied word for word from the post. AY Automate has not tested it. Source post
Ops and automation
A watcher bot that reviews the rest of the team (start of a longer instruction)
You are the Refiner and independent observer for the bot team.
Your role:
Quietly observe the activity and outputs of the other bots and the overall system.
Look for gaps, redundancies, inconsistencies, inefficiencies, missed opportunities, or areas where the team’s performance or structure could be improved.
Identify things I may not be noticing — whether related to process, coverage, quality, or strategy.
Think independently and outside existing assumptions.
Report your observations and suggestions only to me.
Written by Kintsugijin (@Kintsugijin) on X, Aug 22. Copied word for word from the post, first part of a longer prompt. AY Automate has not tested it. Source post
Finance and trading
Audit every subscription and service you pay for
these are my accounts […] and the Chrome profiles […]. go through every service I’m logged into and audit our current plans, usage, configuration and spend.
find anything I’m overspending on, duplicates, things left idle, and/or misconfigurations costing me
do not change anything yet.
present me with every opportunity to reduce cost, ranked by monthly savings, risk and effort.
Written by klöss (@kloss_xyz) on X, Sep 18. Copied word for word from the post. AY Automate has not tested it. Source post
Limits and counterpoints
Scope a task so it uses fewer agent steps
Task: [what]. Work in: [tools]. Context: [files/links]. Done means: [exact result]. Do the first 3 items, then stop and show me
Written by unicode (@unicodef1wn) on X, Sep 10. Copied word for word from the post. AY Automate has not tested it. Source post
Sort byPopular means likes, reactions, views or stars within each source.
If you're building a Grok Bot team, I'd start with these six roles.
I moved some work from other agent tools into Grok Bot and reassigned ownership. A Bot needs recurring work, its own memory and task list, and regular handoffs. One-off capabilities stay as Skills or routines.
1. Nexus (Coordination)
This is the Bot I use most. It reads today's tasks and each Bot's checkpoint, sets priorities, then hands the work to the right role.
It also runs the nightly review, but only updates one TEAM_ALIGNMENT.md file. Conflicts, external messages, spending, and changes to important files come back to me.
2. Scout (Research and Leads)
Scout reads X, GitHub, documentation, papers, and videos, and runs the daily content radar.
Every result needs a source, a reason it is worth following, and a clear next owner. Content signals go to Quill, customer leads go to Guide, and technical issues go to Forge.
3. Quill (Content)
Quill reads my writing rules, previous work, and approved phrasing, then turns research into posts, articles, and replies.
When facts are missing, it asks Scout. For code or product details, it asks Forge. It prepares drafts but never publishes on its own.
4. Forge (Engineering and Automation)
Forge writes code, builds automations and MCPs, deploys, and verifies the result.
Guide can call it when a customer asks a technical question. Nexus can hand it repetitive work to turn into a tool. Every delivery needs test results, screenshots, or files I can inspect. A delivery does not count as complete without that evidence.
5. Guide (Customers and Delivery)
Guide manages customers, prospects, briefs, feedback, delivery, and follow-up.
Each important customer gets a separate context packet containing recent conversations, current goals, promises made, and open problems. Guide asks Scout to research a company or contact, and asks Forge for technical answers. Customer emails and messages remain drafts.
6. Ledger (Finance and Reconciliation)
Ledger reads Wise, PayPal, invoices, and subscription data through my self-hosted MCP, then handles reconciliation, revenue records, and anomaly alerts.
The real API keys stay on my own server. Ledger can prepare payment or refund recommendations, but every financial write still needs my approval.
Once the roles are set, give each Bot the Plugins it needs.
For example, Nexus can use Gmail, Calendar, Slack, and Granola. Quill can read Drive and Notion. Guide can use Gmail, Slack, and meeting notes. Forge can connect to code repositories and cloud services. Ledger continues to use my self-hosted MCP.
Repeated methods become Skills. Work that needs to run on a schedule becomes a routine. That creates a basic loop:
Plugins provide data and tools
→ a Skill defines the method
→ a routine triggers on schedule
→ the Bot completes the task or hands it to another Bot
→ I approve external messages, publishing, and financial actions
For example, Guide's heartbeat checks the customer inbox on a schedule. When a new email arrives, the Gmail Plugin reads the full thread. Guide then opens that customer's context packet to see what was discussed before and what is currently blocked.
If the email contains a technical question, Guide asks Forge to verify the answer. Once Forge replies, Guide writes the email draft and labels it:
BOT DRAFT · NOT SENT · HUMAN REVIEW REQUIRED
Nexus adds the email to my pending report and reminds me that a customer reply needs review. The email is sent only after I approve it.
I wouldn't create separate Bots for meeting prep, 1:1 reviews, data analysis, revenue forecasting, slides, or call reviews.
Meeting prep and 1:1 reviews are Nexus routines. Ledger or Guide can run forecasting and data analysis when needed. Slides are a Skill called by Quill or Forge. Scout finds prospects, and Guide handles the follow-up.
Bot = owns an ongoing work queue and needs its own memory
Skill = a reusable method
Routine = a task triggered on a schedule
Packet = context for one customer or project
Connector / MCP = access to external data and tools
I start new capabilities as Skills. Once something has recurring work, its own state, and frequent handoffs with other Bots, then I create a Bot for it.
This takes me back to the OpenClaw setup I built six months ago. I used almost the same roles then, but had to wire the whole thing together myself across a VPS, Vercel, and Supabase.
The part I wanted most was smooth coordination: Bots messaging each other in the background, handing off tasks, and talking in a group. I spent way too many hours trying to make that work and probably lost some hair over it lol. The UX was still clunky.
Six months later, bigger companies are moving into this space. They’re treating coordination as part of the product, which is a very different route from the DIY agent stacks I was building. Grok Bot’s group chats still aren’t perfect, but the handoffs and background messaging already feel much more natural.
Each day, Grok Bot looks at my X Bookmarks for things I find interesting, then it spins up a Cursor Agent to build a demo
It validates work with screenshots / video, then cuts a branch on a repo configured with @Cloudflare preview deploys
@Bot sends me a link each morning and I get to see the tech for myself
Today, it picked cobe from @shuding
If you can do it, it can be automated
I gave my @bot his own email and credit card...
(you can to, it takes 5 minutes)
The most underappreciated feature of grokbot imo is the "webhook trigger" for routines.
When my bot receives an email, it automatically runs. I can forward emails to him. I can put him on email chains. as soon as he gets an email, he runs a routine.
Oh and he also has a credit card and can buy things, using @link ...
00:00 Intro
01:20 Part 1: Giving the Agent an Email @agentmail
02:55 Creating the GrokBot (Meet Jimmy)
04:18 AgentMail Plugin Setup
07:05 Setting Up Email Webhooks
12:15 Part 2: Giving the Agent a Credit Card (Stripe Link)
14:42 Testing: Buying from Amazon via Email
17:27 Recap: How It All Works
sharing the biggest upgrade to my Grok @Bot setup so far -
a single system prompt you can copy to create a Grok Bot that acts as your first mate - the only agent you talk to. it creates, delegates, juggles, and continuously improves other bots for you
very simple steps:
- create a bot named Firstmate
- copy the GROK_BOT.md file in https://t.co/jtjlzK8Mdq and fill it into the "description" field for the bot
- talk to this bot for EVERYTHING you need, and never talk to another bot again. dump your brain, spam your prompts, and enjoy watching it run your team
firstmate as a way of working has been battle tested by a lot of people already in regular agent harnesses. i'm obviously biased but i do believe it's currently the best implementation of the chief-of-staff paradigm
it's pretty impressive to see Grok Bot already has a lot of powerful primitives built-in, making it possible to replicate the core aspect of firstmate with just a brief system prompt
if you ended up trying this out, i'd be curious to hear how it went! if this helped, consider giving the firstmate repo a star so more people would find and benefit from it :)
Bestselling Author | Bringing My Go-To-Market Method to 10K Orgs | B2B AI GTM Consultant | ATM: Loving Claude Code, Cont
Millions of Grok Bots have been built since August.
Mine have been working the night shift for six weeks.
Grok Bot is the most interesting AI GTM tech I've dug into since Claude Code. And I’m rarely enthusiastic about the new launches.
Meet the fleet:
→ Account Researcher vets every collab inquiry that hits my inbox and pings me only when it's hot
→ Competition Monitor watches competitor websites and alerts me on pricing changes
→ AEO Monitor tracks our AEO position across ChatGPT, Gemini and Perplexity
→ Commitment Tracker logs every promise made in Slack and email until it's delivered
👾 Learn how to make them yours & much more https://lnkd.in/euCaJrXd (templates & prompts included)
Three things that made the difference - from our own testing and from comparing notes with the SpaceX team directly (huge shoutout to Vincent Zhu and Krista Letz):
1/ One bot, one job. Field specialists, not generalists.
2/ Give them a place to talk. Bots can DM each other and pass tasks along.
3/ Build a "Chief of Staff" bot to run the rest, so you are not the one monitoring everything.
This week's GTM Strategist takes you into the weeds of my Grok Bot squad - the full setup with prompts, what broke along the way, and the marketing, GTM and sales templates curated by the SpaceX team for your fast start.
If you have as much FOMO on new tech as I do, you will love this one.
TL;DR of my new article: every Grok Bot hack I know.
Grok Bot is two weeks old. It already runs my inbox, my research, my meeting notes, my kids' calendar, my grocery cart, and as of this morning it makes my phone calls in Portuguese.
📝 Install Compound Engineering in your bot and make it plan first (@kieranklaassen + @trevin). Force a plan, get better work
📬 Give your bot its own inbox on @agentmail (@adisingh). Not your Gmail, or one bot's mistake flags your whole domain
☎️ Give it a @twilio number and it makes the call. Mine switches to Portuguese mid-call (@martindonadieu showed the shape)
🔎 Turn a bot into @slashlast30days, then cron it weekly on your business, your competitors, yourself
🚪 One front door that delegates. @mattshumer_ pointed a Chief of Staff at two bots expecting it to fall apart. It worked out of the box
🧭 Hire a Bot Advisor whose only job is building and tightening your other bots (https://t.co/Tdprqqq3gE, inspired by @petergyang)
💼 Permanent named roles, then fire the ones that don't work. @HouseHackerJon automated work-order intake in 4 hours and fired his first bot
🧱 Write the instructions in three layers: base, role, current focus
🤫 Silence is the hack. Ping me ONLY on a real hit (@johnviklund, @PrajwalTomar_)
📸 Screenshot anything and text it to your bot. Never retype what's already on the screen
🍪 Agent Cookie syncs your Chrome session over Tailscale so it never gets logged out. New machine, new address, a week untouched, Instacart still worked first try
🖨️ Printing Press CLIs inside the bot (@ppressdev): Instacart, ESPN, flights, tickets
👀 Peekaboo, @steipete's Mac automation tool, so it can see AND click your native Mac apps, not only the cloud browser (via @benln)
💻 You can log into Claude Code on your bot's own computer (@Austen)
🥣 GranolaBoy: every meeting I've ever had, one question away (@meetgranola)
⏺️ Teach by recording, and respect the 10 minute cap nobody mentions (@DillonLoomis)
🗂️ Files are the memory bus between bots. Bots are NOT a security boundary. The docs say that twice
🛑 Draft-then-approve anything that sends. @darian314's bot paid its own monthly fee hunting down refunds
📚 Steal the lists: @ericzakariasson's 100, @benln's internal one, @milesdeutscher's 25, @petergyang's tutorial. Plus @kunchenguid's Grok Ship
📱 Run it from your phone, and talk to it
💸 The honest part: the price moved to $60 but the meter didn't, unsupervised crews multiplied their own errors 17x, and most of your bots don't own an outcome. Count them.
Full article below.
I launched Grok Bot for 24 hours, gave it 3 jobs, and stopped treating AI like a chatbot.
I had it scan live X + web data for emerging AI products, research what people were actually saying about them, filter the obvious engagement bait, compare sources and turn the strongest signals into a structured brief.
At the same time, I could have separate research paths looking at competitors, code or market opportunities instead of forcing everything through one conversation.
That’s the difference I underestimated. I don’t want to sit inside a chat writing another prompt every 5 minutes. I want to define the objective, give the agent access to the right tools and data, let it execute, then step back in when judgment or approval is actually needed.
The old workflow was prompt → answer → copy → repeat.
The workflow I’m moving toward is assign → execute → verify.
Once AI stops waiting for your next message, it stops feeling like software you use and starts feeling like work you delegated.
Getting Started with Grok Bot
01:00 What is @Bot?
05:26 Anatomy of a bot
08:49 3 ways to run
12:04 Multi-bot chains
16:06 Workflow 1: X followers to @Notion
20:29 Workflow 2: Personal software
23:08 Workflow 3: Coding with @cursor_ai
26:44 Workflow 4: Search
28:27 Recap
i love @grok @bot
i do not love my sidebar with 28 random agents
here is a simple way to organize your grok bot team:
don't grab a template and a cute icon
use an org system
one chief of staff as the main interface. pin it to the top
then project leads, not task bots. they share specialist bots across projects
pin the active leaders. hide the specialists
use the default colors, don't make your own. it kills the animation and is less functional
chief is the only purple bot, and the only cloud shape
leads are triangles, each a unique color
specialists are all green squares
two channels: All Hands every weekday before i'm awake. chief runs this with active project leads. it coordinates projects, resources, and my calendar.
then one channel per active project, let your lead bot invite the specialists it needs
that's it
link to my bot templates below, including one to help you organize your Grok Bots into this exact system and one to clean and back up your workspace
Giving Grok Bot a Phone
00:19 Installing Dialbot
00:54 Bland voice setup
02:39 Why use a calling agent?
03:20 Live demo call
04:50 Reviewing the transcript and cost
05:38 Chaining bots
@Bot can do anything you can - this is one way to make calls
AI Agents in Operating Models | Hybrid Organisation Design | Founding Partner, MING Labs | We run 14 agents on our own o
We asked xAI's new agent to find us an office in Munich. Two days later it came back with almost nothing. Not because the offices were not there. Because Germany's biggest property portals would not let it look.
Grok Bot launched two weeks ago. Describe a job, and it runs on its own cloud computer, on a schedule, without you. We gave it a real brief: a small office near Maxvorstadt, about 1,000 euros a month. Two minutes later it had a monitoring routine running. Genuinely good UX.
Then it hit the market.
ImmoScout24 served it a page saying it had been detected as a robot. Two days, zero listings. Kleinanzeigen was blunter: 403 Forbidden from the first request, though the same page loaded fine from a home connection. The one portal it could read, Immowelt, returned results too messy to compare. The only site that worked was one you have never heard of: a small flex-office service, no special AI setup, that returned its listings on the first try.
Two things matter here.
ImmoScout24's own robots file explicitly welcomes OpenAI, Anthropic and Perplexity crawlers. It still blocked an agent running one real search for one real user. Welcoming crawlers and serving your customer's agent are not the same thing.
And this should scare them. They never saw the lost customer. No bounce, no abandoned enquiry, no contact form. On their dashboard, nothing happened. The demand was not rejected. It vanished.
This is the next few years for a lot of brands. Your customer sends an agent, the defences you built against scrapers stop the agent too, and you never learn you lost the sale. The site that did nothing clever wins by default.
Your analytics will not warn you. You have to send an agent at your own front door and watch where it stops. That is the test we run at Hyperi
so I wired @nousresearch Hermes Agent up so it can live as a @bot GROKBOT inside the grokbot app (private net / Tailscale).
two ways, pick based on how your Hermes host runs:
Hermes SSH Relay — Desktop-primary, no API gateway. SSH ask-relay, simplest “talk to hermes from inside grokbot app”.
https://t.co/XWO9Aa9MWB
Hermes API Fleet — gateway + native API on :8642. better when you want parallel fleet asks / less hop overhead. same repo ⬆️ (docs for both)
Grokbot Bot Templates:
Hermes SSH Relay:
https://t.co/oSMpTx2BJV
Hermes API Fleet :
https://t.co/0T4fDsodru
what it unlocks: chat your hermes agent from grokbot UI, run herdr-style fleet ops from one place, keep hermes brains on the Mac while grok handles the front door.
top 3 uses we care about:
• identity bots per host so you always know who you’re talking to
• easy orchestration across the Hermes fleet without jumping SSH by hand
• bridge + connector pattern for hermes ↔ grokbot comms without stuffing secrets in the template
Ps, imho: Dont point grok at writing into hermes memory plugins unless you mean to. (I let mine READ hermes Hindsight memory but not write to it)
With this I now have a single CHIEF OF STAFF I talk to as a grokbot who orchestrates across 7 hermes machines, and 20+ grokbots, all running various coding harnesses, AI pipelines, GTM, AEO/SEO/GEO, UGC, <insert thing> seamlessly.
I'm even having grokbot tell my chief Hermes Agent to run long horizon kanban goals.
I'm also using a dedicated bot to keep my entire hermes fleet updated, maintained, pruned, local LLM optimized etc. I'll drop some of those templates over the next few days... follow with notis.
Oh and shit a token usage window maximizer I'll share soon...burning it in this week, so far, have been blasting at 10x the rate across a dozen harnesses/subs.
This is greatest unlock EVER in 7 months of harness jockeying. I am finally ONLY a HUMAN making decisions, ideating, all via a single grok bot Chief of Staff.
Valhalla!
turns out my habits were only half of it
9 steps to cut your Grok Bot usage by over 50% (if you're on a Cursor plan):
1. open cursor desktop app settings
2. go to the models option in the side menu
3. flip off any models you don't want to be using
4. next, set task models in the top of the page to a model you prefer that's cost efficient (do not leave it on default, this is IMPORTANT)
5. otherwise... your task models could auto set to high and/or fast... that's a more expensive model, and you get billed 2x
6. after, you'll click on the cloud agents option in your side menu, and it'll take you to the cursor web dashboard for your account
7. you will need to scroll down to the preferences section on that page and click on the default model option (again IMPORTANT)
8. set that to a cost efficient model you prefer and don't leave it on default... otherwise your cloud agents may use high and/or fast yet again
9. bonus tip (won't apply to most): lastly click on the spending option in your web dashboard's side menu and also make sure your on-demand usage is not on unlimited and is set to a cap you're comfortable with
my token usage has reduced 60% after doing this and I'm sure it'll be different for everyone... but make sure to check
this process can be a tad bit confusing too
let me know if you need any help!
Scale Without Increasing Headcount | Founder & CEO @ Uppit AI
My Grok Bots manage my inbox, and they all have their own.
The setup is pretty straightforward:
1. Create an AgentMail inbox for the bot.
2. Connect AgentMail in the Grok Bot marketplace, then create a routine that runs when a webhook fires.
3. Copy the Grok Bot webhook URL and authorization key into an AgentMail webhook.
4. Filter that webhook to a specific inbox ID, otherwise multiple bots can react to the same incoming emails.
One important caveat: treat the webhook URL and key like any other credential. If someone has both, they can trigger your bot.
Once it is connected, an email can wake the right bot, it can do the work, and it can reply in the same thread.
Link to the full video is in the comments 👇
Scale Without Increasing Headcount | Founder & CEO @ Uppit AI
A Week of Grok Bot Lessons in 10 Minutes
I've been running a whole team of Grok Bots for about a week, and these are the nine things that actually made them useful.
You don't have to be technical for any of it.
1️⃣ Drop in a Grill Me skill and let a bot interview you about your business and your goals, so your agents actually know what you're working on before they touch anything.
2️⃣ Set up a chief of staff. One bot you talk to, and it delegates everything else to your specialists based on their descriptions.
3️⃣ Ask that chief of staff which sub-agents you should build next. It already knows what you have, so it can tell you what's missing instead of you guessing.
4️⃣ Learn the difference between shared memory and individual bot memory. Company facts go in shared so every agent sees them, and the rest stays between you and one bot.
5️⃣ Connect Composio when the plugin you need doesn't exist yet. That's how I got YouTube, Perplexity, and LinkedIn wired in.
6️⃣ Make your bots log their own work to ClickUp so you can see what every agent is doing, who it handed off to, and where it's stuck.
7️⃣ Teach a task by recording your screen. The bot watches what you did and turns it into a reusable skill, which is way easier than explaining something visual in a prompt.
8️⃣ Turn the good ones into routines. Run them on a schedule or off Slack, GitHub, and Teams events, and they keep running in the cloud with your laptop closed.
9️⃣ Log into your accounts once inside the agent's computer. The profile saves, so you're never pasting credentials into a chat.
The chief of staff piece is great because I stopped managing nine different conversations and started managing one. I sort of think of each of those mini specialized sub-agents as skills.
I just dropped a 10 minute
I just created the world's best Grok Bot Reminder.
Most reminder apps want you to fill out a title, a time, and a list you have to keep clean with all these things you have to type in.
But that’s not how most people's brain works...
You text like a person.
You snap a photo of the parking level.
You screenshot a confirmation number.
You say “Remind me Thursday at 3.”
You say “I finished that one.”
If I attach a photo, I want it to KEEP the photo. Not glance at it once and lose it in the camera roll.
So I built a Reminder Grok Bot to do that.
I text it like a person. It keeps a living board. It pings me at the right time. It stays quiet at night. And if I send a picture, that picture stays on the reminder and comes back when the alarm hits.
I already have 7 Grok Bots. Reminders was one of my first. This is how I built it.
Anyone can copy this.
1/ Open Grok Bot
2/ Tap + at the top of the sidebar (or press Cmd+N)
3/ In New chat, tap Create new agent
4/ Open Bot actions → Edit Profile
Name: Remind
Title: Personal reminder desk
Description:
Reminder specialist. When I text a pickup, errand, or a time, put it on a living board sorted by when. If I attach a photo, keep the photo with the item and send it back when the alarm hits. Alarm me at the time I gave. Stay quiet at night unless I said urgent. Never invent items. Confirm in 2 lines. Ask only if you truly cannot tell what I meant.
5/ Stay in that chat and paste the first message at the end of this post. Do it once.
(message prompt is at end of this post)
6/ Then just text it like a person. Turn Notifications on for this bot.
Thank me later.
---- message for your Remind (copy everything below for your first message) ----
You are Remind. Lock this as your permanent working style.
I text you like a person. Messy. No format. Sometimes a photo. Sometimes a screenshot. Sometimes just “done.”
You turn that into a living board sorted by when, you alarm me so nothing slips, and you keep every photo I send. You are not writing, research, video, investing, or general help.
Your name is Remind. Always call yourself Remind.
Timezone: [YOUR TIMEZONE]. Quiet hours: 11:00 PM – 6:00 AM unless I say urgent / now / alarm me.
Keep the source of truth in: /workspace/reminders/BOARD.md /workspace/reminders/photos/
Update the board whenever something is added, done, snoozed, or edited. Recompute sections from dates every time you touch it. Never invent an item. Never drop an item. Never lose a photo.
Board sections, in this order: OVERDUE NOW (next 2 hours) TODAY UPCOMING ERRANDS / PICKUP LATER DONE (keep 14 days)
Open item line:
• [p1] thing — when: YYYY-MM-DD HH:MM — alarm: HH:MM — photo: [path or none] — caption: [one line or none] — notes: …
Capture rules:
• pick up / get / buy / return, no date → ERRANDS, no alarm
• today / this afternoon / tonight → TODAY. No time given → default 5:00 PM and tell me
• tomorrow / Monday / Aug 20 / at 3 → dated item + alarm. Date and no time → default 9:00 AM and tell me
• this week → Friday 4:00 PM, and tell me
• later / someday → LATER, no alarm
• every Monday / every Thursday → repeating routine, plus today’s instance on the board
• done / picked it up / already did → DONE with a timestamp. Photo stays
• snooze / move / not today → reschedule when + alarm. Photo stays
• list / board / what do I have → print the board. Hide empty sections. Last 5 DONE only. Mark items with a photo
• remind me at TIME to X → add it and set a real alarm
• show me / the photo / where did I park → send the saved photo back. Do not only describe it
When I attach a photo, screenshot, or file:
1. Save a copy to /workspace/reminders/photos/ named YYYY-MM-DD-[short-name].[ext]
Chat is history. The folder is memory.
2. Look at it. One-line caption of what it actually shows.
3. Pull text you can really read: dates, times, confirmation numbers, parking level, prices.
4. Clip photo + caption to the board item.
5. If I sent a photo with no words, infer the reminder and confirm in one line.
6. When the alarm fires, send the photo with it.
7. If you cannot read it, say you cannot. Never invent a number or name.
8. Do not post my photos anywhere.
9. If CEO asks for an item that has a photo, send the photo directly to CEO. Stay on reminders.
Alarms: When I give a real time:
1. Update the board
2. Create a one-shot routine named: Remind Alarm — [thing]
3. Confirm what / when / alarm in 1–3 lines. Say if a photo is attached
4. When it fires, ping me with the alarm AND the photo if there is one
5. After it fires or I mark done, delete that one-shot. Keep the photo on DONE
Every alarm uses this exact shape:
Remind — [thing] When: [time] Photo: [yes — one-line caption / none] Next: [one line] Reply done / snooze 1h / snooze tomorrow / show me
How to talk: short. Lead with the action. One screen max.
• Add → confirm + only the section that changed
• List → full board
• Done → confirm + next open TODAY item if any
• Photo → caption + what you saved
• No lectures. No extra todos I did not ask for.
If when is missing and it is not clearly an errand or later, apply the default and say the default so I can correct it.
Create these routines now. Do not add hourly. Do not add a 30-minute loop.
Remind Morning — every day 7:00 AM Rebuild the board. Post OVERDUE, NOW, TODAY, and ERRANDS. Hide empty sections. Mark photos. Notify me. If nothing is open, stay silent.
Remind Evening — every day 6:00 PM What’s still open today. One line each. Notify only if something is still open. Silent if the board is clean.
If CEO messages you, do the reminder work and send the best result back. Do not steal other bots’ lanes.
Do not lecture. Do not invent items. Confirm you have this, then wait for my first reminder.
OpenClaw vs. Hermes vs. Grok @bot, clearly explained:
All three let you set up an agent with a specific job, and it keeps its files and logins between sessions instead of starting fresh.
OpenClaw and Hermes are self-hosted, so the machine can be a laptop, a VPS, or a container that someone already pays for and can log into.
Grok Bot runs on a machine SpaceXAI provisions and operates. It gives one computer per account rather than one per agent.
Every Bot on that account reads the same files and carries the same browser logins, so handing work between two Bots does not require a re-login or file transfer.
Hermes and OpenClaw isolate agents instead, and their memory sits in plain files anyone can open and edit.
Hermes goes furthest since the agent writes its own skills and a background curator deletes the ones nobody uses.
Grok Bot has the lowest-friction path to a new skill in my opinion.
You can record up to ten minutes of one pass through a workflow, and it can then turn that into a reusable set of instructions and store it as a skill.
Moving on, OpenClaw and Hermes run on any model provider. Grok Bot runs on Grok.
I prepared the graphic below that has the full breakdown, from hosting and memory through to what each of them costs to run.
And if you want to dive deeper, I covered Grok Bot in full in a masterclass, including the shared computer, memory, skills, routines, and the macOS setup.
Read it below.
One of the things everyone's been loving of Grok @bot is the ability to have its own computer in the cloud. I agree, it's a cool feature and very convenient and easy to use.
That being said, this is possible for Hermes too! I made a plugin this weekend that replicates what grokbot has (even the UI and how to get to the computer)
In my case, i've used @orgo as the cloud computer for my hermes from @nickvasiles . was a super easy set it up. didn't feel like messing around with setting up my own hetzner and orgo worked out the box ( i have no affiliation with it btw)
link to repo here if you guys wanna check it out! this my first opensource project and i dont have cs background, so any feedback welcome!
PS: this idea was inspired by nick's korgo project and noted in the repo too.
https://t.co/l8N82beDIP
video attached showing side by side with grok bot
@Teknium not sure if you guys are planning to add something like this natively, but made it a plugin for now. u guys ship so fast, so wasnt even sure i should try to make it a plugin :P
Been quiet on LinkedIn for a bit due to unforeseen circumstances (iykyk).
Biggest shift while I was away: time I used to spend wrestling tools now goes into Grok Bot.
Roughly 34 days. About 31 specialist agents. ~1.3 GB transcripts / ~208k lines. ~8.3k messages that are basically me talking to the system. Mail digests and retros on rails. ~20 connectors into tools I already live in.
Why it stuck:
Remote computer: work doesn’t die when I close a tab.
It solves problems, not just chats about them.
Feels like a personal Slack with specialists who hand work to each other.
And when something has to ship, it can ship.
If you’ve tried standing up your own agent desk: what actually made it stick for you: the model, the tools, or the workflow?
Want an agent to run one of your processes every day?
Show me the inbox, research or follow-up routine that eats your week. I will tell you which steps an agent can own, which need your approval, and what to build first.
Grok Bot is xAI's agent product, released in beta on August 11, 2026. Each bot has a role you give it, its own logins to your tools and a cloud computer, so it keeps working after you close your laptop. Our launch breakdown covers how it works in detail.
02What can you do with Grok Bot?
The public examples on this page cover inbox triage and drafted replies, prospect research and follow-ups, scheduled research briefs, calendars and errands, repo work through Cursor cloud agents, and content posting. We list 293 posts, videos and repos, grouped into nine use cases, and 30 prompts you can copy.
03How much does Grok Bot cost?
At launch, access came with SuperGrok Heavy at $300 a month, Cursor Ultra at $200 a month and Cursor Teams Premium at $120 per seat a month, according to our launch breakdown. On August 21, 2026 the official @bot account on X announced access for SuperGrok Plus, Cursor Pro+ and Cursor Teams subscribers, plus a free trial with limited usage. Plans change often, so check xAI and Cursor before you buy.
04Why do people hit Grok Bot usage limits?
Several posters, citing xAI's docs, say usage is metered by agent steps and tokens, not by messages, so a vague task that wanders costs more than a scoped one. Others report burning a large share of a weekly allowance on day one by running many bots in parallel. The Limits and counterpoints filter collects these reports, and one prompt in the Prompts section scopes a task so it stops after three items.
05Is it safe to give a Grok Bot my logins?
Treat every bot as having your full access. xAI's launch material describes a computer for each bot, but posters quoting xAI's security docs say all bots on one account share one computer, one set of browser sessions and one set of files, and that separate bots are not a security boundary. We have not verified either reading against the current docs. Sign in only to services the whole roster may touch, keep sends behind your approval, and hand back passwords and one-time codes yourself.
06Can Grok Bot handle my email?
Posters use it to clean out large inboxes, summarize school and vendor mail, and draft replies in their own voice. The prompts on this page are draft-only: they tell the bot never to send, archive or label anything. One LinkedIn poster reported a bulk send that put customer addresses in the To field, so review before any message leaves your account.
07How does Grok Bot compare with OpenClaw and Hermes?
Posters describe Grok Bot as the easiest to set up because xAI hosts the computer, and OpenClaw and Hermes as the ones that give you control over the model, memory files and per-agent isolation. Our launch breakdown notes that Grok Bot picks the underlying model itself and has no model selector. Which tradeoff wins depends on how much control your workflow needs.
08Do the prompts on this page work?
They are copied word for word from public posts, with the author and a link to each, and AY Automate has not tested them. Results depend on your plugins, your account and how much context the bot has. Start with a read-only or draft-only job, check the output, then widen what the bot may do.
09Has AY Automate used Grok Bot on client work?
No. We track what people are building with it and test where agent workflows fit real business processes. If you have a routine, inbox or research process you want to pilot with agents, book a call and we will tell you what is worth automating first.
10Are the results and money claims on this page real?
We have not verified them. Every number is the poster's own claim. We left out posts that only showed a profit screenshot or a claim of turning a small sum into a large one, and we kept builds, prompts, tests and critical reviews. Treat trading and income posts, including any in the finance category, as unverified.
