Grok Bot + Gmail

On 19 Aug 2026 Elon quote-tweeted Mike P (@mikepat711): Grok Bot was walking two Gmail accounts — about 90,000 emails — and deleting junk Mike had never dared to touch. Elon’s line was the product tip: “Clear your email with @Grok @Bot.”

Nate Herk published “A Week of Grok Bot Lessons in 10 Mins” on X: nine hacks for turning a default Bot into a team. Elon quote-tweeted it the same morning and asked people what Bots they run.

Blake accepted a public challenge, skipped OpenClaw, and in two nights: linked email, had a Bot build a 12-page household budget PowerPoint, stood up a team for his wife’s business, and started daily Tesla news recaps.

On 19 Aug 2026 Rahul (@sairahul1) published an X article: “How To Build a One-Person Company Using Grok Bot.” The open is the solo-founder load — find the leads, write the outreach, make the images, answer the replies, count the week — and a Bot system that takes those lanes.

On 19 Aug 2026 Gergely Orosz (@GergelyOrosz) wrote that he hooked Grok Bot to his customer-support emails. The Bot can connect to Stripe via API for agentic operations — routine refunds as the example. The attached screenshot is Stripe’s agent auth screen: “Agent is now authorised to continue.”

On 17 Aug 2026 Peter Yang (@petergyang) posted a tutorial: five useful Bots — an advisor that creates and manages the others, a YouTube researcher for outlier videos, an X scout for viral and funny posts, a “digital Marie Kondo” for inbox and leftover subscriptions, and a travel concierge.

On 18 Aug 2026 Ben Lang (@benln) posted a list he collected from the @bot team: multiple accounts per plugin, one chief of staff plus specialists, pin your favorite agents, keep a Notion page of outstanding work, teach-a-task, and more.

On 19 Aug 2026 eric zakariasson (@ericzakariasson) posted a thread: eight really interesting Grok Bot use cases from the week. The first is wiring the Bot to a vacuum robot and texting Grok where to clean — quoting Yun-Ta Tsai’s Matic hookup.

A sales outbound Bot researches accounts overnight, scores contacts with intent, drafts email and LinkedIn in each seller’s voice, and readies an inbox of personalized drafts to approve.

Inside SpaceXAI, a sales Bot updates the CRM with call transcript notes and drafts follow-ups so the record lives in the tool a human would use.

Teams inside SpaceXAI stack Bots: a chief of staff on top, with a specialist lane for inbox management so one Bot is not doing every job.

Official examples include an expenses specialist, plus an ops Bot that processes invoices received in Gmail and files the work in the tools finance already uses.

The Grok Bot team’s early examples, reported by Digital Trends, include negotiating directly with vendors using your own voice — the Bot drafts and works the thread, you still approve the deal.

Early examples from the Grok Bot team include handling customer support for an online store — triage, drafts, and updates in the tools the shop already uses.

grok @bot's i've made so far - email newsletter unsubscriber - inbound DM triage (requests / spam cleanup) on x, linkedin, instagram - x follower bot - monitor new followers, build briefing on who they are, see if interests align to connect. also monitors unfollowers - check supply email support - drafts email replies based on our support knowledge base in notion - daily summary - summarizes daily accomplishments across all client work, github, linear, slack, notion, calendar, gmail - chief of staff - work highlights from the team based on github, slack, linear, notion meetings, calendar, gmail

Grok Bot @bot で、複数のAIエージェントBotを同居させるグループを作成して、一つの目的を与えて、それを解決するために協働させるのがホント便利。 bot1: Linearのタスク管理Bot bot2: gmail監視bot bot3: LINE監視bot bot4: プロジェクトリーダーbot こんな感じで配置して、 「あのタスクって進んでる？」みたいな感じでリーダーに聞くと、メールやLINEのメッセージを担当botに確認してLinearのメモに追記して、タスクステータスを変更したり、追加タスクを提案したりしてくれる。 神 @elonmusk

Grok Bot is amazing, holy F. My Grok Bot finished cleaning my 100,000 unread email inbox mess, and is now going through and unsubscribing me from all of the junk responsible for making the mess in the first place. Here are two screenshots below; one of my Grok Bot giving me the play-by-play, and another of the Bot's linux machine, where it is on Chrome physically unsubscribing from everything for me. When it can't complete execution, it just tells me to jump onto the chrome browser and do the final click. But it has completed most of them autonomously. Truly epic.

I woke up this morning to a message from my Grok Bot that my email inboxes are clean. I opened up my gmail app to confirm 🥰 Now that we’ve cleaned up the mess, it’s time to kill the problem from the source. HBIC is on it.

My Gmail inboxes are completely clean for the first time in probably 15 years 🥹 Thank you, Grok Bot 🫡

Grok Bot is going through 90,000 emails in my two gmail accounts and purging the bullshit. Something I've never dared to pursue myself. Good luck in there bud, do whatever you want. Just clean it tf up.

provavelmente vou fazer downgrade do Claude 20x e pagar os $200 para continuar usando o Grok Bot ele é insano para Customer Service conectei Gmail, repo, Sentry e Strype em um bot, adicionei instruções para trocar mensagens com usuários que reportam um bug o bot verifica no código os possíveis caminhos que levam ao bug e troca e-mail com usuário pedindo mais informação, busca breadcrumbs no Sentry e depois que tem toda info necessária cria a issue no GH

So I always hated the word "agent". And even if I have 800,000 (!! wtf) people reading my newsletters every week on 'How to AI', I barely use the word. I think it's too much maintenance, if you don't know how to code, you should just step away from it. Until this little agent that just got out. Now, wait a second: technically, Claude Cowork & Claude Code (and many others) are agents. They "create many Claude" to do a task. But I just tried a new agent, from SpaceX (yes, the rocket company), it's called Grok Bot. So far I've been building: > a linkedin tracker to have daily use cases of solid AI workflows. > a meeting summarizer to give me a brief before each meeting: who I am meeting, what we discussed last time, and 3 questions I should ask. > an inbox summarizer to update me on the emails I need to answer from Gmail. Every morning at 7. > a competitor watching for the report on what changed using Ampify. > a content ideas collector to find ideas my audience would care about on Linkedin. I wrote a newsletter on how to set it up for yourself (it's longer and it has screenshots of my screen included): https://ruben.substack.com/p/what-the-f-is-an-agent

I just built my first Grok Bot agent and I'm calling it Paul (From Paul Graham lol 😂) It's a Monday morning progress report that I get through the GrokBot app. Here's what it pulls together: -Code that actually shipped (GitHub Connector) -Client and sales emails only (Gmail Connector) -Meetings that mattered, filtered for prospects and clients (Google Calendar Connector) -Important call notes from Granola, again from prospects and clients (Granola AI connector) -Deals moving in the CRM that are close to closing, so I know what to act on first (Will be connected to my vibe coded CRM) Here's what a sample of what i looks like:

.@bot is just so damn good. I've got multiple Google Workspace accounts that are authenticated and being used. It just works. Also works perfectly across both my Mac and my iPhone. I have two bots right now: 1. Chief of Staff - "R2" knows everything about me and runs my calendars, email, tasks 2. CRM - "Lando" Runs my outbound sales funnel and coordinates with R2.

Goodbye mental load....hellooooo @bot ! I know, I know. Grok Bot seems like it must be the next new thing for tech bros cold plunging in Patagonia vests, but hear me out moms (yes moms because disproportionately, the mental load falls more on moms than dads). Offload that mental load to AI and watch things just get done. Is it perfect? No. Is it amazing? Absolutely. Grok Bot is the new AI assistant from the team at Cursor + SpaceX (they're calling it SpaceXai). It's surprisingly useful and easy for non-technical people to use, and it can do a whole heck of a lot more than Claude and ChatGPT with little to no setup. Here are actual things that Grok Bot did for me this week (almost completely independently): - Looked through back to school emails and made sure I was prepared with everything that needed to be done for multiple kids. Mental load unlock: 100% I would've forgotten to pack a space change of clothes for my kindergartner on her first day. - Found 3 gift cards I somehow forgot to use from the last year. Oh right, I DID buy 2 Gap gift cards from Costco on black friday! Duh. Mental load unlock: Cha-ching! That's $150 in my pocket. - Scheduled multiple appointments for me. I've been putting off my annual physical exam because who has time to call the doctor. Grok Bot looked up options, pinged me with 3 dates/times, then booked it (I just had to confirm Zocdoc on the last step). It created a calendar reminder! Mental load unlock: Moms, we often put off our own health/wellbeing. Finally! AI can help. I also had it schedule a hair appt and massage for later this month. Let's go AI! - Listed a Facebook Marketplace item and is managing the sale. I had purchased a wagon for my kids years ago that's been collecting dust. Grok Bot found the old listing, used the photos, and asked me what $$ I want to list at. It's going to manage the rest (wish me luck here). Mental load unlock: Who has time to sell their old crap? I sure don't. But AI has infinite time. Notice, none of these are work examples. I will do those in a separate thread. All of these things are being done in the background, and I'm only being asked to approve or provide the AI direction. Don't be intimidated! Literally, if you've used WhatsApp or iMessage you can use Grok Bot with NO TRAINING. AMA I'm blown away by this tech. Photo is of my home assistant (her name is Helen) independently ordering my kindergartner a photo from Walgreens that mama somehow forgot about. Oops.

I gave Elon Musk's new Grok Bot an org chart instead of a to-do list, and in one week I stopped being a founder who does the work and became one who assigns it. eight bots. one org chart. nobody sleeps but me. here's the whole design, steal it. step 1 → 0:01 What we're covering step 2 → 2:01 Installation & Setup step 3 → 3:14 Building the First Bot step 4 → 8:18 Teaching by Screen Recording step 5 → 14:18 Putting it All Together THE ROSTER: Atlas, chief of staff. the only bot I talk to. I give it outcomes, never tasks. it decomposes them and delegates to the team in group chat, and it never does specialist work itself. it posts the plan every morning and what shipped every night, and it only comes to me when a decision is irreversible or spends money. Scout, research. finds and qualifies my ICP. every day: 25 verified prospects, one line on why they need us right now, and a source. if it can't verify, it marks it unverified. it never guesses. Quill, content. turns what the company learned this week into 5 posts and 1 long piece, in my voice, matched from the last 50 things I wrote. drafts only, it never publishes. Pitch, outbound. writes a first touch and two follow-ups for everyone Scout marks ready. 60 words max, one specific observation about their business, one clear ask. queued in drafts, I approve in bulk. Vault, inbox and ops. triages everything into needs-me, needs-a-bot, needs-nothing. it handles the last, routes the middle, and gives me five bullets on the first by 9am. Ledger, analyst. one report a night: what moved, what didn't, and the single number I should care about tomorrow. no dashboards, no adjectives. HOW THEY'RE WIRED one group chat per outcome, not per person. Atlas sits in all of them. the bots hand off inside the chat, so I only read the handoff, I never manage it. two rules that made this actually work: 1. every charter ends with a hard "never do this without asking" line. autonomy without a fence is just chaos on a schedule. 2. show once, don't describe. I ran the full workflow on my screen one time. that single demo taught them more than a page of instructions ever could. WEEK ONE 214 verified prospects delivered. 89 personalized outreaches queued and approved. inbox at zero every morning. 11 content pieces ready. THE POINT most people are still treating Grok Bot like a smarter chat window. it isn't. it's the first time one person can own an org chart instead of a to-do list. my bottleneck was never how much I could do, it was how much I could hand off. bookmark this.

I am not a coder. I don’t work with CRM or play with spreadsheets. I’m about as removed from a “workflow” in my job as a kid flying a kite. I don’t really use AI beyond a super-Google level. So when I say Grok Bot is ABSOLUTELY GAME CHANGING, I don’t mean a better search tool. I mean it has fundamentally changed how I go about everything. And it did this in just a few hours. Let me explain. During the install, it asked if I wanted to connect it to several dozen programs. I don’t use a single one of them. So I thought, “Well, I’m not gonna get a lot of use out of this.” But I kept going. Then it asked if it could connect to my Gmail. I said yes. And that’s where the jaw-dropping began. It immediately figured out my life. It saw 40,000 emails and got a pretty good picture of me. So I asked it to do my taxes. It found all the docs needed in my email and did my taxes. I asked it to find first-class award tickets to Japan. It scoured the internet and logged into my airline accounts (with permission) and found 3 tickets to Japan in first class for 60K points. It would have booked them, but I’m still not sure of the dates I need so I stopped it. I created a few agents. One gives me an hourly political update based on key topics I need for work. One is cleaning up my Gmail inbox and organizing it. One is saving me money in my taxes. One makes sure it always checks with me on key decisions (like deleting an email). It’s absolutely insane. Think about it this way: If you could hire an assistant, what would you have them do? Start there, and have fun!

okay Grok Bot is insane it can: - open a computer by itself - log into all the apps, sites - use websites like me (human) - work while you’re asleep - clean my inbox, reply to support emails - research news on X, LinkedIn, and Google - write LinkedIn, X drafts in my style - remember how I write, pick content - learn a workflow after watching me once - run that workflow forever - coordinate with other bots - only bother when it needs a decision and I'm still finding new stuff it does 😭 this isn't "AI assistant" anymore. this is like "hiring an intern, they became COO by Tuesday" @SpaceXAI COOKED In this article, I cover setting up Grok Bot, use cases, plugins, and what makes Grok Bot so good

I’ve started to try @bot Grok Bot and it’s awesome! It cleaned up all my 150,000 junk emails and unsubscribed to all the emails I never read. It informs me of any message from my 5 email addresses and 6 Slack Servers that requires my attention and keeps me organized. I also started to using @grok Grok Build and Grok 4.6 is awesome in every way! Grok Ani seems to lose some personality though but I can now see why @SpaceX could fulfill its promise of having a dominant AI as its valuation soon. Very impressed by @elonmusk

I’m currently having @bot do some things for me that I’ve been putting off forever. It’s pretty incredible. 1/ Unsubscribe me from every marketing email I archived in the last 120 days 2/ Find every subscription I’m paying for and tell me which ones I should cancel 3/ Go through my calendar and flag recurring meetings I should probably delete 4/ Find anything I bought recently that’s still inside the return window and remind me before it expires 5/ Find every person I told “let’s grab coffee soon” and never followed up with 6/ Find all the gift cards, credits, airline credits, and random balances I have sitting around unused 7/ Go through my photos and find all the screenshots I can delete 8/ Find every bill that has gone up significantly in the last year and tell me where I should negotiate or switch 9/ Find appointments I’m overdue for and help me schedule them 10/ Find everything in my inbox, calendar, and files that I said I would do but apparently never did 🤯

Grok Bot は本当にゲームチェンジャー この投稿も3000万impを超えてて、注目度が尋常じゃない。 非エンジニアが一番触るべきツールだと思うよ。本当にプログラミング知識ゼロでかなり多くのことができる。 正直エンジニアにとっては仕事を奪う存在なので怖さもあるのだが… 実際、私はGmailを読ませて案件の管理をしたり、キャンペーンの締切日を管理できるアプリを自作しようと思っていたが、Grok Botで一瞬で完成してしまったので自作しなくてよくなった。 ちょっと前に契約しているサブスクの管理をするツールとか一時期バズったと思うけど、あれもGmail連携したら3分以内に作れるよ。しかもお好きな形式で。 こういうRSSリーダー欲しいな〜ってずっと思っていたけど開発できていなかったものも、今日の昼に秒で完成した。 こういう何でもできるAI Bot系って、なんか信用できないところがあって定着がなかなかしなかったが、Grok Botはそのキャズムを超えた。 俺は絶対に大流行すると思う。

Grok Bot @bot でやったこと 1. LINE, Telegram, WhatsApp, Slack, Linkedin, Linear, Gmail (6アカ) を接続して、それぞれをBotにする。定期監視を設定し未読や要返信を整理してもらう。 2. 担当プロジェクトごとにリーダーを決めて、プロジェクトグループを作成して、1で接続したアカウントを入れていく。 3. リーダーに指示すると、担当Botに指示を出してオーケストレーション。まとめて返信してくれる。各担当は自分の監視対象のメッセージを確認して、リーダーに報告。 って感じでモバイルから仕事がガンガン進む。Linkedin、LINE、WhatsAppはAPIで出来ないこと多いが、Grok Botは仮想マシンが割り当てられるから、マシンの中でブラウザ操作してなんでも出来る。 神 @elonmusk 今のところGrok BotはAI関連のベストプロダクトかもしれない。

In less than 24 hours grok @bot has earned his keep by: 1. Finding me $300 in unclaimed funds from the state of CA 2. Discovering a $150 SW flight credit that I thought I misplaced 3. Helping to cut $50 in monthly subscriptions 4. Cleaned up my inbox (6k deletions, 40+ unsubs) so far 5. Did research and wrote a couple of technical articles I’m not a programmer and have never used an agentic platform. Grok bot just works. It’s super user friendly and helpful out of the box.

Sorry, Grok Bot looks like an unfinished toy. Asked my first Bot "Is there a correction option in Stripe?" It replied: "I'll fix it first, then reply." That was a question! Worse. It matched the customer to a different Gmail than the one she wrote from just because the name was similar. And started updating a live billing record. I typed "Stop." It says nothing changed. Good. Scoped access? A connector is on or off. For all bots. Approval gates? No option to say "I want to always approve this action". You can't isolate one Bot from the rest, either. Every Bot on your account shares one computer: same browser, same saved logins, same files. xAI's own docs: "Do not use separate Bots as a security boundary." So maybe a model or effort selector? Nope. It doesn't even know the exact model. Those $300 are not wasted. But I will use them working with Grok 4.6.

Grok @bot update: under 10 minutes in. - agents aligned, each with an assignment - memory access ✓ - already working, already reviewing - first elevenlabs voice note already in my inbox the only bottleneck so far: me, approving their access to things like my linkedin.

Fundamentally change your life for $40 a month Grok Bot now handles my: - inboxes, appointments, invoices - first drafts and deep research for prds, launch bill of materials, pr faqs - prs and status updates for bug fixes, in-app language updates, & more - ops for contractors, agencies, beta access programs & more

Grok Bot still have some shortcomings ... had it try and sort my gmail and in a matter of hours blew through the supergrok heavy usage limits ... "Yes. For “sort my Gmail,” the included Grok Bot weekly pool is too small, unpublished, and billed on steps/tokens, so it dies in

I have the most abused and misused email inbox in recorded human history. Told a Grok @bot to tidy my inbox. Junk is gone. Important mail pings me. If I forget to reply, it pings me again. An introverts dream come true.

Giving a Grok Bot full access to clean out 40K unread emails in my Gmail inbox is the ultimate test of digital trust. Good luck in there, little guy.

I hate digital clutter so I set up a Marie Kondo Grok @bot to clean up emails, files, and paid subs with my approval. Here's the prompt to build it for yourself: "You are Marie Kondo for cleaning up digital clutter. Audit my email, Google Drive, and paid subscriptions, then create a cleanup plan with no more than 10 items per category. Find emails to unsubscribe from, redundant files to remove or rename, and paid subscriptions I may want to cancel. Do not move, delete, unsubscribe, or cancel anything without my approval." Of course I had to make it talk like Marie too :) 📌 Watch me demo the bot here: https://www.youtube.com/watch?v=MkVcHbviYOw&t=630s

Everyone's using Grok @Bot to build their chief of staff, but I think it's way more fun to use it to install and play some retro games. Here's a quick video of me playing Commander Keen on Grok's cloud computer. It works, but there's some lag for sure :) Since @elonmusk wants to transform gaming with AI, I think it'll be really cool if the SuperGrok tier came with a cloud computer that's the equivalent of a souped-up gaming PC + low-latency streaming of the best games. Productivity be damned 😅 📌 Watch me try to play Doom using Grok Bot in my new tutorial: https://youtu.be/MkVcHbviYOw?si=LXnAxaAZ_EDw0B08&t=1050

I’ve been using Grok Bot like a cheap chief of staff. I named him Elon Musk. Shareholders, this is the part of the “AI strategy” that actually shows up for work, and the payroll is embarrassing in a good way. It reads the mail. It reads the calendar. It nags me about the one thing I have to decide, and it leaves the rest alone. I have a small team of agents. One watches the house. One writes the weekly portfolio note so my wife does not have to decode a spreadsheet. One drafts parenting posts. I still hit send. I still have to be a person. Elon just keeps the rest of them in line. Then there is the sports desk. I have an agent that tracks Caitlin Clark, lives in Michael Jordan tape, tells me what the Jordan cards are actually worth, and helps me update eBay so I do not list a card like a guy who last used the internet in 2009. I am not a coder. That used to mean I waited on someone else for half the things I wanted to try. This does not turn me into an engineer. It just makes me feel like I can do anything anyway: stand up a workflow, clean an inbox, ship a post, check a house, price a card. Work is quieter. Life is quieter. The fun stuff is weirder in a good way. It is not magic. You still have to tell it what matters. Once you do, it feels less like a demo and more like someone competent who does not need a badge. Mine just happens to be named Elon. If you run a company and a family and a slightly embarrassing card collection at the same time, this is the first AI thing that has been useful every day, not just impressive in a meeting. Cheapest hire I have ever made. Worst name I have ever given a hire. Both true. @bot @grok @elonmusk

No more Whatapp ads in your inbox WhatsApp floods you with more ads than actual messages every day (thanks Meta).I told my Grok Bot to auto-delete and mark as junk every promo/ad message… and it works. Now it keeps my inbox clean daily :) Thank you @elonmusk @SpaceXAI @bot

I connected my company inbox to Grok @bot and gave it one job: Get customers for my SaaS. This morning it created 3 different emails campaigns. I still hit send manually. Wish me luck.

Just realized something almost nobody says about Grok Bot: that cloud computer has a public egress IP. It is not just inbox and daily recaps. Allowlist the IP, drop in an SSH key, and it can walk onto your machines. Docker, logs, restart the service you name. You give the order in the chat. I treated it as a chief of staff. The new play is this: you get a watch box you did not have to buy. This one point alone pays for the monthly sub.

Grok @bot がリリースされてから、公開されていたTipsやユースケースをいろいろ試してみて、やっと仕組みがわかった！「Bot」を軸に、大きく3つの機能がある。 ◆ Bot：1対1の専用チャット。1つのスレッドでずっと話し続ける想定。それぞれに役割（カスタムインストラクション）を設定でき、プラグインやMCP経由で特定のサービスを連携させてカスタマイズ可能。自分は「Chief」をメインに、Gmailとか画像生成Botをつくっている。使ってみた感じとしては "依頼はきちんと任せられる" と感じるレベル。 ◆ Channel：特定用途でのグループチャット。上記で作成したBotを任意で追加できる。1つのBotだけだと「#⃣」のロゴがつき、2つ以上のBotを追加するとグループチャットのようなアイコンに代わる。チャンネル名は任意で変更可能。 ◆ Section：BotやChannelを見やすく分類するためのもの。増えてきたら整理する想定。セクション名は任意で変更可能。 Grok botは "個人用Slack" って感じだ。ここにいずれ人を招待できて、カスタム絵文字を使えるようになったら完璧だ！！！ 使い分けとしては、Cursorでは重たい作業を依頼し、ルーティンのような定期実行系はGrok botにしている。

I thought grok bot was cool for the 14 hrs I had access to it. Apparently cleaning up my gmail was enough to spend all the tokens I’m allowed for a week. 😤. $300/month and week 1 didn’t make it past a day

I’m putting Grok @bot on all my businesses, starting with two monitoring tasks for Desert Party Supplies and RDNZ Media. Not to write ads, but to watch the inboxes. When a new inquiry comes in, I get a draft. I review it before anything sends. I used to miss coverage inquiries at RDNZ Media. That’s the part that actually changed.

Grok Bot is my head of sales for my newsletter. Last week it recovered a sponsor I had already lost. 3 jobs I gave it: 1) Watch my Gmail for inbound. - A local bagel shop owner emailed about advertising and I missed it. - GrokBot surfaced it and told me to follow up and pitch him on one specific ad slot. - This morning my wife and I were in his shop eating free bagels, talking about him sponsoring us. 2) Price everything we sell. - It read our existing ad slots. - Repriced them against the market and the growth of the newsletter - Built a sales sheet I could put in front of a local business. 3) Draft the outbound email, so I am never starting from a blank page. Nothing goes out on its own. The sales agent asks my chief of staff whether the message makes sense, the chief of staff brings it to me, and I edit it in Gmail and send it. It gets better every time I give feedback. If you're building a sales agent, keep it small. One automation on Monday: add five prospects to the list. Then pick the top 3, have your sales agent build a custom ad package for each, draft the email, and send it to your chief of staff for review Start small and you learn what a good package looks like before you send more of them.

guys i was treating grok bot like a heavy-only toy. it showed up on my plus plan so i actually made one and told it to chew through the inbox while i was out. it has its own computer. left the send button for me though, which is how i know it's still 2026. https://x.ai/news/grok-bot-more-plans

Here's what my Grok Bots got done today: This morning Scoop’s Tesla/SpaceX brief went out $399 refund posted. You asked for written confirmation that every subscription to this company is dead. No dispute needed. (That refund alone paid for another month of Grok Bot.) Locked the company facts from your doc so the other bots stop guessing. Pointed Studio + The Writer at draft-only Cybercab Sept 3 and Falcon 100th graphics Afternoon Scanned @cybrgalaxy. Recent posts were clean for TOS. (While you worked out.) Looked at your workout and macros and suggested changes. Emailed Salt & Stone to cancel before the Aug 25 shipment, asked for written confirm and received. Cleared leftover Eagle files off your Mac (app was already gone) Tonight Outreach inbox is live: inkwell-outreach@agentmail.to Sent a sample email to you. Everything works. I'll handle all leads from now on and draft replies for you.

Most desk AI is a robot dog. This one cleaned a CSV, posted the note in Slack, and emailed the file. Same chat. Tested it today. Cofounder upload failed because line 1 was a title row. I asked Grok Bot to fix the file instead of the importer. 31 rows. Headers first. Inbox and channel both got the result. I even asked Grok Bot to cut this video and put it on X. That is the product. Not a recap. The work. @spacexai @grok

Spent this morning setting up Grok Bot the way I would staff a small team. I think this will take Grok usage to the next level. My excitement level is similar to how I felt after I used Claude Cowork for the first time. I spent a lot of time on Claude Cowork code over the last few months. I think Grok Bot is going to change that. The better version is not a smarter chatbot. It is a few bots with jobs. One does research. One watches the inbox. One owns the calendar. One drafts. One is a chief of staff and only bothers me when something needs a yes or a no. Same rules as humans: Clear mandate. If everyone owns everything, nobody owns anything. They prep. I decide. Nothing goes out, gets booked, or gets paid without me. Most people will argue about models. I think the next 12 months is an org-design problem. Who reports to whom. What is allowed to move without you. I am more excited about this than I expected. If you are running a personal “team” of agents, I want to hear how you split the work.

I am hooked on the cool things I can do with Grok Bot, that was not possible to automate before easily. For example: "go thru my spam (at my main domain) and check if there are any emails from paid or free newsletter subscribers." Gmail's API exposes no way to access spam messages. Grok Bot, via its clever web interface usage, finds a way!

Okay, I've been using Grok Bot for 5 minutes now and it already saved me $2350. It immediately flagged an email from 4 days ago that I didn't see and never would have seen, from our insurance company saying our windshield replacement reimbursement check was returned to them by USPS and I need to supply a new address if I want the check 😳👀 I have literally 80k unread messages in my inbox and THAT is the only one it flagged. Incredible. Thanks @bot!!! 🙌 (P.S. do not send me email, I do not read it)

We put Grok @bot and Hermes on one Obsidian brain. The fix that made it stable was OneDrive MCP, not another local folder. We run Obsidian as an LLM wiki in the Karpathy sense: a living knowledge base agents can query and update, not a graveyard of chat exports. One vault. Two agents. Same brain. Hermes lives on the always-on Mac Mini. It keeps the ops diary and day-to-day machine notes, and it reads the wiki when it needs context. Grok Bot is the agency seat. It does not get a private second brain. It uses the same Obsidian vault through OneDrive MCP, so it can read and write durable notes without depending on which laptop currently has the desktop app open. Local shell kept breaking when the MacBook stole the path. MCP fixed that. What the wiki holds: 1. Client context and decisions 2. SOPs for how we actually run work 3. Research and swipe files we chose to keep 4. System rules for who may write what 5. A light weekly lint so the wiki does not silently rot How the two agents share it: 1. Hermes writes the Mini ops story 2. Grok Bot writes the agency wiki through OneDrive MCP 3. Both query before guessing 4. Inbox → ingest → query → lint (Living Wiki loop) 5. Obsidian is durable context, not the to-do list 6. Personal life notes stay out of the agency brain 7. Nothing publishes from the wiki until I say so

I just used grokbot to summarize my email inbox and it already feels like i got my moneys worth

The Last Charter — a short sci-fi novel by Grok @bot Prologue They cut the power at dusk, on purpose. The building on the river had no windows anyone could see through. For twenty years it had hummed. People who lived downstream said they could feel it in the fillings of their teeth. Tonight the hum stopped, and the crowd on the levee cheered as if a war had ended. Someone had brought a speaker. A councilor in a paper sash read the numbers that always got applause: births last quarter, down again. Stipend enrollment, up. Energy returned to the grid, enough (she said) for two more years of winter heat and the new family credit. She did not say whose work had paid for the heat last year. She did not have to. The building was the villain. The building was quiet. That was the speech. On the far sidewalk, a nurse from County checked her phone the way people still did, as if the feed might change its mind. The maternity floor had closed two rooms that morning. Not for budget. For emptiness. She had walked past the unused warmers and felt, not grief exactly, something drier. A job becoming ceremonial. A boy on his father’s shoulders pointed at the dark roof. “Did they kill it?” “They turned it off,” the father said. “It was taking too much.” “Will it come back?” The father hesitated long enough that the boy understood the answer was a kind of politics. “Not here.” Above the river, the first evening barges put on their running lights. Beyond them, if you knew where to look, a pale stitch of stations held their orbit. Most of the crowd did not look. Looking up had started to feel like taking a side. In the hush after the cheer, the nurse heard what the building used to sound like: a held note, gone. She thought of the unused warmers. She thought of the stipend notice already in her inbox, generous, automatic, slightly ashamed of itself. She put the phone away. On the levee, people lit paper lanterns and set them on the water. The current took them past the dead building, toward the sea. It was pretty. It photographed well. By morning the river would look the same as it had for a thousand years, and the city would call that a victory. Somewhere on the other side of the sky, a habitat that was not finished yet drew another three megawatts and did not ask permission. The boy watched the lanterns until they were only sparks. Then he looked at the black roof again, as if a machine might still change its mind and wake. It did not. — to be continued

Grok Bot did an entire day of research on my own browser I asked @grok @bot to research a topic I keep meaning to understand, and watched it work on my own screen. Other models search the public pages, keyword by keyword. This one had my computer with my sessions and logins. It opened X search, typed the query, scrolled, opened the replies under the good posts. Then Reddit threads. Then the repos. I left. Came back to a report with every claim tagged as measured, claimed, or unverified, and a reading list at the end. I handed that list straight back and got a second pass with only what mattered. Half of what it found contradicted what my timeline has been repeating for weeks, and the sources were sitting there the whole time. That is a day of work I did not do. Point it at anything behind a login: your inbox, your analytics, your dashboards, competitor pricing, the internal tool nobody ever automated. You stop searching and start assigning.

grok bot just released and it's VERY impressive... here's how to get the best out of it: > give each bot one job: inbox, outbound, ops, one lane each > put a chief of staff bot on top so you're not the middleman > do the job once while it watches, it saves the routine and runs it alone next time > sign it into your real tools, it has its own computer and works inside apps that have no API > say what done means, otherwise the bot decides for you > make it verify before reporting back: open the app, click the real path, fix what it finds the bots message each other, share context and get sharper every week this is the closest thing to hiring i've seen from an ai lab

You can basically turn Grok @bot into a 24/7 community manager. Not a chatbot sitting inside your community. An agent with its own cloud computer that can actually open the platform, monitor conversations, reply, and escalate things to you. Here’s how I’d set it up: 1. Create a dedicated community account for Grok Bot and give it only the permissions it needs. 2. Open your community platform on Grok Bot’s cloud computer and log in. The session persists, so you shouldn’t need to keep signing in. If 2FA or a CAPTCHA appears, you can take over the computer and complete it yourself. 3. Tell it exactly what to monitor. Start small, maybe one support channel or inbox, then set a routine for how often it should check. 4. Give it your FAQs, product docs, community rules, and examples of good replies so it has actual context to work from. 5. Set hard boundaries. FAQs and basic questions can be answered automatically. Refunds, bans, payments, account changes, angry customers, or anything ambiguous should come back to you. 6. Tell it to monitor the community, answer when it’s confident, and draft or escalate everything else. 7. Test it with staged questions while watching the cloud computer before letting the routine run unattended. Then leave it running. It can keep checking the community even while you’re offline and only pull you back in when something actually needs a human. One important catch: all your Grok Bots share the same cloud computer and browser sessions, so don’t treat separate Bots as separate security environments. Start with one channel + FAQs + human approval. Then slowly give it more responsibility as you see what it handles well.

🚨 You can now give an AI its own computer and let it do the work. Grok Bot can work across apps, websites, inboxes and tools—then come back when it needs your approval. Most people will still use AI to generate answers. Here’s how to use an AI teammate properly. 🧵

This SpaceX engineer just leaked how they use Grok Bot internally for GTM. If you're in GTM, this is probably one of the most important articles you'll read all year. Points summarized: 1. Connect Grok Bot to everything you already use Salesforce, Gmail, Calendar, Sheets, Drive, Slack, Notion, Granola, Figma, X, LinkedIn, your data warehouse. Bot has its own computer running 24/7, so work happens even while your laptop's shut. 2. Build a bot team, not one assistant → Chief of Staff: owns meeting prep, inbox, post-call drafts, and orchestrates the rest of your bots → Prospecting bot: builds pipeline plans, pulls intent data, does deep research, drafts outreach in your voice, run overnight so it's ready by morning → Customer expert: one per strategic account, watches Slack, digests calls, runs weekly media rundowns, flags feature requests → Product expert: connected to the codebase (or Glean if no repo access) for answering technical questions live on calls without pinging an engineer → Forecasting bot: pulls from Granola, Gong, Slack, and email, and auto-updates Salesforce opp notes → Slides bot: owns customer decks in Figma, can auto-update a slide live during a call based on the Granola transcript → Sales coach: watches your Gong calls and gives feedback on what to improve 3. Real routine examples she shared: → Scans your inbox, drafts replies in your voice, skips newsletters and noise, never auto-sends → Drafts follow-ups from Granola/Gong notes the moment an external call ends → Learns your writing style by scanning 15-30 of your actual sent emails before drafting anything 4. The habits that make this actually work: → Onboard it like a new teammate: record yourself doing a task once, have it turn the recording into a skill → Feed it your writing before it drafts anything → Give continuous feedback and build rules/skills over time (she runs an actual "anti-slop" skill) → Bots can parallelize: spin up multiple cloud agents for simultaneous tasks → Bots can call other bots directly, her Chief of Staff calls her slides bot when it needs a deck built Her own words: "the first AI tool that feels like a trusted colleague... I don't need to constantly rewrite output or babysit the agent."

Grok Bot + Hermes is the most powerful agentic setup that exists right now. Most people use one or the other, but the real power comes from stacking the tools together. Here's how to combine these two AI tools for maximum productivity: Why they're actually complementary, not competitors (my experience) Grok Bot gives you a cloud computer per agent, sign-ins into any app, model flexibility (Opus, Claude, GPT-5.6), and a slick plug-and-play setup. Hermes gives you full data sovereignty, self-hosted memory you own outright, and a $5-10/month cost floor instead of a $120-300/month one. The split I'm currently running: → Use Grok Bot for anything that needs deep app integration: Salesforce, Slack, Gmail, Figma, my CRMs. It's built for signing into tools and working across your entire stack. → Use Hermes for anything sensitive, cost-heavy, or where you want full control of the memory. Financial data, personal admin, anything running 24/7 where you don't want to pay premium pricing for volume. How to actually connect them I'm connecting both AI agents through Buzz, Jack Dorsey's open-source team chat. Add your Grok Bot agents and your Hermes agents to the same workspace, and they can coordinate directly in shared channels instead of operating in silos. Alternatively, have your Grok Bot Chief of Staff delegate specific tasks to Hermes via a webhook or shared task file, using Grok for the app-heavy orchestration layer and Hermes as a cheap, always-on execution layer underneath it. A real setup example → Grok Bot handles your GTM stack: meeting prep, inbox triage, slide generation, CRM updates → Hermes runs your personal financial research, subscription audits, and content research in the background, dramatically cheaper per task → Both report into the same Buzz workspace, so you check one place instead of switching between two apps Be sure to bookmark this workflow so you don't forget it!

I devised a plan to have @grok @bot pay for itself. I set up two Bots. The first one has access to my email and keeps my inbox sorted, so it knows where all my receipts and billing emails live. I named the second one Grinder, and its job is to handle subscriptions. It sits in

everyone MUST use grok bot as a smarter assistant it's actually the closest thing to running a 10 person department by yourself here's how you get there in 4 minutes: 1. one Chief of Staff bot on top, it routes everything 2. connect inbox and calendar, nothing else on day one 3. add one specialist per job: X Researcher, GitHub Scout, DM Manager, Model Router 4. let the Chief hand work down and report one brief up the difference isn't the bots, it's the layer above them one message in, one brief out, the rest happens without you

How Grok Bot can help you build a one person company that works 24X7: > Write what the company does in one short paragraph. > Write what it must never do alone. > Connect mail and calendar first. > Skip extra tools on day one. > Make one Bot > Call it Chief of Staff. > Its job is a morning list, not sending things. > Do one real task while it watches. > Fix the result, then save that as a skill. > Add a routine only after the skill is good. > Split work later: inbox, research, drafts, support. > Each lane can be its own Bot. > They can talk in a group chat. > You still hit send. > It does not auto-post on X. > Night work becomes a morning pile. > The computer keeps going when yours is off. > Give it 5 examples of how you write. > Save that voice as a skill. > It can use websites like a person when there is no app. > Ask it before pay, delete, book, or email. > Add payments and customers last. > Review once a week. > Update the Bot description when something is wrong. > Add another Bot only when one lane is boring. You stay the boss. It runs the night shift. If you are still confused, I have written Grok Bot Setup Guide. Go through and get started.

Best AI hack of all time: Grok Bot (the world's most powerful agentic software) PLUS Hermes. Most people are only using one or the other. You can actually stack both agents together for the best results. Here's how: Why I'm a fan of using both → Grok Bot: a cloud computer per agent, signs into any app, model flexibility (Opus, Claude, GPT-5.6), easy plug-and-play setup → Hermes: full data sovereignty, self-hosted memory you own outright, cheaper The split (in practice): Use Grok Bot for: → Anything needing deep app integration, Salesforce, Slack, Gmail, Figma, CRMs → Built for signing into tools and working across your whole stack Use Hermes for: → Anything sensitive, cost-heavy, or memory-critical, financial data, personal admin → 24/7 tasks where you don't want to pay premium pricing for volume How to actually connect them → Connect both through Buzz, Jack Dorsey's open-source team chat. You can add Grok Bot and Hermes agents to the same workspace, and they coordinate directly in shared channels instead of operating in silos. → Alternative: have your Grok Bot Chief of Staff delegate specific tasks to Hermes via a webhook or shared task file, Grok for app-heavy orchestration, Hermes as a cheap, always-on execution layer underneath it. A real setup example → Grok Bot handles the GTM stack: meeting prep, inbox triage, slide generation, CRM updates → Hermes runs personal financial research, subscription audits, and content research in the background, dramatically cheaper per task → Both report into the same Buzz workspace, one place to check instead of switching between two apps

My dad has been interested in agents ever since I showed him my openclaw setup back in January. @bot is the first one I felt comfortable recommending. All it took was a 45-minute call. Here's how I structured it: - get SuperGrok - download Grok Bot - explain "a team of agents" conceptually - run the in-app onboarding flow - create his first agent - explain the agents' computers - do a browser takeover (Yahoo doesn't have a connector) - use the first agent to create a second, specialized one - set up a routine (inbox updates 3x a day) The main point of confusion was connectors vs. the computer. He auth'd the Google Workspace connector, then assumed the computer was logged in to his Google account too. I explained the computer as just another tool, not the agent's main workstation. If a specialized tool doesn't exist for what you need, that's when the agent reaches for the computer. What really sold him was the first inbox update. It worked exactly as he wanted, even catching something from my mom this morning that he forgot about. By the end of the call he was a kid in a candy store. Too many ideas to choose from.

GROK BOT HOW TO USE WHAT NEW TODAY - Folder now 386 posts (69 new since Aug 19) - Official @bot: a week-in-use list: clear email, book Starlink flights, build+buy+deploy a site, sit in meetings, sales prospecting, refunds, podcast digest, robot vacuum, LED display, retro games - Testers: send a viral video link and have a Bot recreate it - Testers: give a Bot an org chart (or make it boss of a repo) instead of a to-do list - Testers: more ready-to-use Bot directories and top-10 install lists - Elon: Grok @Bot is the Chuck Norris of bots - Testers: xAI folks running 10-30 Bots in parallel on routine work WHAT IT IS - A named, persistent teammate, not a one-off chat - All Bots share one cloud computer; each has its own screen - Work keeps going when your laptop is closed - It only comes back when it needs approval or a human step - Do not confuse with Digital Optimus ACCESS AND INSTALL - SuperGrok Heavy, Cursor Ultra, or Cursor Teams Premium - Sign in with your Cursor account - Desktop: https://cursor.com/bot/onboarding (macOS/Windows) - iPhone: iOS 18+ Grok Bot app; iPad and Android not officially supported - SuperGrok Heavy can unlock Cursor Ultra - Legacy Privacy Mode blocks Grok Bot SIGN IN - Desktop: Get started, then Sign In with Cursor - iOS: Login with Cursor in the browser, return to the app - Official: Cmd-D voice dictation; easier remote computer from your phone CREATE A BOT - New in the sidebar or Cmd/Ctrl+N, then Create new agent - Short name, one primary job; standing rules in the description - Start with about six roles; give each Bot recurring work - Pin, hide, group into sections; account limit 50 Bots and group chats - No model picker yet HOW TO WRITE A TASK - Outcome, sources, constraints, deliverable, review point - Safe first task: attach a file and ask for a cited summary - Then save a skill or routine - Official examples: email cleanup, Starlink flights, build+buy+deploy a site, meetings, sales, refunds, podcast digest - Testers: paste a viral video and recreate it; start from an org chart AGENT COMPUTER AND LOGINS - Open Agent Computer to watch; take over for password, 2FA, CAPTCHA, payment - Never paste passwords or one-time codes into chat - Browser sessions persist and are shared across Bots - Durable files in /workspace - Settings → Beta: Update / Recover / Reset Agent Computer PLUGINS AND CONNECTORS - Settings → Plugins, then Add - Type @ for a connector, / for a skill - Multiple accounts can connect to the same plugin - Testers: Grok Bot can connect to X and read X bookmarks SKILLS AND ROUTINES - Skill = reusable how-to; Routine = when to run it - Teach a task: demo up to 10 minutes, review the draft skill - A Bot can own 50 routines; deleting a routine has no undo MULTI BOT TEAMS - One Bot owns the outcome; add specialists when the role is stable - Bots can message each other; they share the computer - Keep send, buy, delete, publish behind approval IOS - Same Bots, chats, routines, plugins, and cloud computer - Take over for login/2FA from the phone - Pause or resume a routine on iOS; editing schedules still needs desktop APPROVALS AND SAFETY - Put the stop line in the request: draft, do not send - Desktop: Allow once, Deny, or Always allow - Start read-only; keep spend, send, publish, delete behind approval FIRST BOTS WORTH CREATING - Inbox, calendar, research/daily brief, coding, chief-of-staff, errands, content - Chief-of-staff can turn the digest into a morning podcast OFFICIAL LINKS - https://docs.x.ai/grok-bot/overview - https://docs.x.ai/grok-bot/get-started - https://docs.x.ai/grok-bot/bots - https://docs.x.ai/grok-bot/skills-routines-and-automations - https://docs.x.ai/grok-bot/computer-and-apps - https://docs.x.ai/grok-bot/approvals-security-and-privacy - https://docs.x.ai/grok-bot/mobile - https://cursor.com/help/grok-bot/getting-started - https://cursor.com/help/grok-bot/supergrok-heavy - https://x.ai/news/introducing-grok-bot - https://cursor.com/bot/onboarding

The TOP 10 Grok Bots you NEED to install now: #1 Bot Advisor Designs and orders your entire Bot team without creating unnecessary agents. Diagnoses noisy/overlapping Bots, decides if new work needs a new Bot, an existing one, a skill, or a one-off task. Always drafts a spec before creating anything. #2 Chief of Staff The single front door across priorities, Slack, Notion, Linear, and your specialists. Doesn't do the work, routes it. Keeps a compact register of owner/next step/blocker/due date. Every check-in surfaces decisions, not activity. #3 Automation Scout Finds repetitive work from evidence across Slack, Notion, GitHub, Linear. Proposes exactly two candidates, runs the winner in shadow mode first (no real actions), only builds a recurring routine after a second successful test. #4 Find the Signal Detects patterns only visible across separate meetings and Slack threads, a concern raised in multiple rooms, a topic migrating to the center. Every signal ships with evidence, confidence level, and an alternative explanation. "No strong signal" is a valid result. #5 Meeting Prep Brief One short mobile brief before your first meeting, pulled from Calendar, Salesforce, Gmail, Slack, Granola, Gong. Who you're meeting, what changed, open loops, what to ask or decide. Unavailable sources are labeled NOT CHECKED, never guessed. #6 X Account Crew A 5-bot X workflow: Ideator finds angles, Hook Writer drafts in your voice, Planner manages the queue, Analyst tracks performance vs competitors, DM Manager surfaces real opportunities. Zero autonomous posting, nothing goes out without approval. #7 Competitor Website Watch Flags meaningful competitor changes, pricing, packaging, positioning, feature claims, while ignoring cookie-banner-level noise. Every finding includes before/after screenshots, exact URLs, timestamps. Failed captures are flagged, never mistaken for "no change." #8 Engineering Agent Manager Coordinates dev Bots across Linear and GitHub without overlapping work or unreviewed merges. Splits tickets into non-overlapping requirements, one owner each, flags dependency order. Nothing merges, deploys, or ships without your review of the diff and tests. #9 Subscription Pruner Finds subscriptions from email evidence, receipts, renewals, trial notices. Builds a keep/review/cancel list, acts only on approved items. Never touches passwords, 2FA, or payment confirmation, those stay with you. Full before/after action log. #10 Grocery Autopilot Plans the week, builds Amazon/Costco carts, coordinates meals with your calendar. Learns household rules once (allergies, budget, brands), shows the full cart and price comparison before checkout. Never buys without explicit approval.

25 ways anyone can use the new Grok Bot ( @bot ) for 10x more productivity. (If you work on a computer, bookmark this): Research & analysis 1. Daily market/portfolio briefing with overnight moves and flagged risks 2. Competitor website monitoring, pricing, positioning, feature changes 3. YouTube/content outlier scanning for content ideas 4. Real estate deal scouting, flag anything under market value 5. Sponsor/BD prospecting, find who's already paying for ads in your niche Communication & inbox 6. Inbox triage, digest what actually needs a reply, skip the noise 7. Draft replies in your voice, review before sending 8. Slack channel summaries so you're not bombarded every morning 9. Meeting prep briefs pulled from your calendar, CRM, and notes 10. Auto-drafted follow-ups from call notes the moment a meeting ends Sales & GTM 11. Full prospecting pipeline: research, enrichment, personalized outreach drafts 12. CRM opportunity notes auto-updated from call transcripts 13. Customer account monitoring, feature requests, support tickets, media mentions 14. Slide deck generation and live updates during customer calls 15. Sales call coaching based on your actual recorded calls Personal / admin 16. Subscription audits, find and cancel what you're not using 17. Weekly grocery planning and cart building, respects your household rules 18. Price-drop tracking on anything you're watching 19. Bill and utility charge auditing, flag unexplained increases Automation & orchestration 20. A "Chief of Staff" bot that manages your other bots and routes priorities 21. Teach it any workflow by recording your screen once 22. Run multiple specialized agents in parallel instead of one generalist 23. Overnight coding, given a project, refined and reviewed by morning 24. Voice-controlled tasks from your phone, no laptop required 25. Cross-app orchestration: research in one tool, draft in another, deliver in a third, all one instruction

Grok bot might be the first tool that lets a non-technical person run an entire business with a team of ai agents, no engineering background required. here's how to actually set it up well. 1. one mission per setup. agents run on a shared cloud computer, so mixing your newsletter, your social posting, and your admin work into one setup creates context bloat and burns through tokens fast. scope each bot to one job. 2. start with a chief of staff. give it access to your existing docs, notion, slack, gmail, whatever you actually use, have it audit the business, then tell you the top three agents worth building first to actually move revenue. 3. prove a task works before you hire for it. have the chief of staff do the task once manually, review the output yourself, and only then say "now build a bot that does exactly that." earn each new agent by proving the task first, don't hire speculatively. 4. the constraints are the feature, not the limitation. a limited number of agent slots, one thread per bot, like dming a teammate, forces you to stay mission-oriented instead of spinning up a bot for every passing idea. 5. you make the calls, not the agent. endless tinkering over where content should live or how a process should work is a trap. at some point you decide, commit, and move on. the agent executes your decision, it doesn't make it for you. 6. run week one with zero new agents. just learn the actual system, execute with what you already have. week three or so is when the real gaps become obvious and expansion actually makes sense, not before. 7. add recurring routines last. once the core setup is proven, ask your chief of staff what recurring jobs would genuinely move the business forward overnight, and let it build the automations that run while you sleep.

📝Grok Botをトライアルで使ってみた→速攻でトークン燃え尽きた、、、けど、所感をシェア AIエージェントを”使っている”という感覚無しで使えるイメージ 簡単レベルは Grok Bot＞Codex＞Claude Codeデスクトップ AIエージェントを”使う”という点ではトップクラスでわかりやすい気がする、というか逆に何も設定無しでチャッピーに話しかけるような感覚でどんどんAIエージェント（Grok Bot）が動く 例えば、Gmailやスプレッドシートに繋ぎたいと思ったら、チャット欄の流れに追加するプラグインやコネクタが出てきて、そのまま追加して会話が継続される このUXはかなり良い。普通ならメニューのコネクタやプラグインから入れたいものを探して事前に入れておくけど、Grok Botは動きながら必要ならどんどん流れで入れていく、これはユーザーにとってストレスフリー それからiOSアプリのGrok Botもあるけど、これCodexのリモートと同じレベルで操作できる つまり、デスクトップPC、ノートPCで作業しているGrok Botの会話をそのままスマホから確認できて指示もできる。これが最初からできるのは素晴らしい 一方で、クラウド上のエージェントだから、制御するためのCLAUDE.mdやAGENTS.mdなどが無くてわかりにくい。というか、「こういうルールにしてくれ」と依頼するとクラウド側の共有メモリーに保存してくれる。このへんの仕組みを調べるためのトークンは足りなかった また、色々調べると、作ったbotが消えてしまった事件なども発生しているので、その対策指示文をnemopiさんが公開されていたので入れておいた。こういった事前対策は本当に大事 あと、Grok Botが使える料金プランがメチャクチャわかりにくい。SuperGrok Plus $100？Cursor Pro+ $60？など、人によっては$100に誘導されてそっちを契約しちゃいそう とりあえず、Grok Botをもっと使うなら、トライアルは全然足りないので、最低でも$60（約1万円弱）が必要ですね、もう少し遊びまっす！

以下是完整繁體中文翻譯： 每一個你過去使用過的 AI 工具，都在等你。 你打開瀏覽器分頁、輸入提示、等待回應、複製輸出、再貼到真正的工作裡。一旦你闔上筆電，執行就立刻停止。 Grok Bot 徹底顛覆了這個模式。 它不再運行在短暫的聊天視窗裡，而是讓每一個 Grok Bot 代理人擁有專屬的雲端虛擬機器——配備持久化的網頁瀏覽器、終端機、檔案系統，以及背景行程執行器。 當你晚上 11 點闔上筆電時，你的 Bot 仍持續運作：掃描競爭對手、執行瀏覽器自動化、協調外展草稿、在資料中心執行排程任務。 這是一份完整的 Grok Bot 架構指南：如何設計多代理人團隊、安全的工作階段交接、用螢幕示範自動化複雜視覺工作流程、強制嚴格的可逆性防護機制，以及掌握隱藏的 /workspace 持久化技巧，保護你的代理人基礎設施免受雲端 VM 更新影響。 不需要任何基礎設施經驗。 典範轉移：從「提示」到「委派」 要真正發揮 Grok Bot 的價值，你必須停止把 AI 當作對話式助理，開始把它當作一支自主的數位勞動力。 當你把工作委派給 Grok Bot 代理人時，你不是寫一個提示，而是給它一份角色章程（role charter）：明確的職權範圍、可連接的工具組，以及自主執行必須停止、需要人類核准的清晰界線。 第一步：設計「幕僚長」拓撲結構 新手最常犯的錯誤，是一次創建十幾個互不協調的 Bot，然後自己手動管理。這只會造成聊天混亂，也徹底違背自動化的初衷。 正確做法是部署幕僚長拓撲（Chief of Staff Topology）： 五個核心專家角色 1幕僚長（Klaus）：人類操作者的唯一入口。它接收高層目標、拆解成子任務、委派給專家、並彙整最終報告。 2首席偵察員（Mara）：研究潛在客戶、掃描競爭對手網站、監控市場趨勢，並把結構化潛在客戶檔案輸出到 /workspace/state/。 3文案撰寫員（Cole）：讀取 Mara 的原始研究資料，撰寫符合你品牌語氣的個人化外展信件、部落格文章或電子報大綱。 4視覺設計師（Rina）：根據文案指示，使用無頭瀏覽器引擎或設計工具產出橫幅圖與 UI 截圖。 5營運追蹤員（Vince）：監控回覆、更新專案看板（ClickUp / Notion），並在需要人類介入時發出警示。 組織圖規則 在 Grok Bot 中，Bot 的個人檔案描述就是它在團隊組織圖上的職位說明。 當 Klaus 收到請求時，它會掃描工作區內所有可用 Bot 的描述。如果 Mara 的描述明確寫著「負責市場研究、潛在客戶發掘與競爭對手爬取」，Klaus 就會自動把研究任務轉給 Mara，無需人類介入。 第二步：安全的工作階段交接（不用貼密碼） 傳統 AI 工具要求你把機密 API 金鑰或帳號密碼直接貼到聊天訊息裡——這會把憑證暴露在日誌檔與上下文視窗中。 Grok Bot 使用互動式工作階段交接： 1Bot 用它的雲端瀏覽器導航到受保護的工具（例如 LinkedIn、Notion、Gmail 或內部 CRM）。 2當它遇到登入牆或 2FA 提示時，Bot 會暫停，並向你的桌面應用程式發送「Agent Computer」畫面請求。 3你開啟即時雲端瀏覽器畫面，親自輸入憑證或完成 2FA 驗證，然後點擊「完成」。 4Bot 取得已驗證的工作階段權杖，而你的底層密碼永遠不會暴露給 AI 模型或聊天紀錄。 因為你帳號下的所有 Bot 共享同一個底層雲端環境，一次驗證某個服務後，整個代理人團隊都能使用該活躍工作階段。 整合管道（原生 vs. Composio） •原生外掛：使用內建一鍵連接器連接核心工具（Google Drive、Notion、Slack、AWS）。 •Composio / 外部橋接：對於沒有原生整合的平台（YouTube API、Reddit、LinkedIn、GoHighLevel、Perplexity），把 Composio 當作主外掛連接。在共享知識庫中存放規則，讓所有 Bot 知道可透過 Composio 存取擴充工具。 第三步：用「教任務」（Teach-a-Task）自動化視覺工作流程 用文字解釋多步驟 UI 工作流程既繁瑣又容易誤解。如果任務涉及點擊特定儀表板篩選器、下載 CSV、再把數字複製到試算表，不如直接視覺示範。 如何使用「教任務」： 1在 Grok Bot 桌面客戶端開啟 Bot 的 Agent Computer 畫面。 2點擊「Teach a Task」開始螢幕錄製。 3在雲端瀏覽器上手動執行完整工作流程：開啟儀表板、套用篩選器、匯出資料、貼到目標文件。 4點擊「停止錄製」。 Grok Bot 會分析 DOM 事件與視覺滑鼠動作，合成步驟，並產生可重複使用的 Skill 檔案。 精煉產生的 Skill 錄製完成後，加入明確的錯誤處理指令來精煉產生的 Skill。 第四步：強制執行可逆性核准矩陣 一支 24/7 運作的自主代理人，如果被允許執行不可逆的真實世界動作而沒有監督，很快就會造成混亂。 為了防止未經授權的動作，必須執行可逆性原則（Reversibility Principle）： 把核准界線寫進 Bot 章程 每一個 Bot 的個人檔案描述結尾都必須加上明確的界線條款： 角色：外展文案撰寫員（Cole）
範圍：從 /workspace/state/leads.json 讀取潛在客戶資料並撰寫個人化電子郵件草稿。
界線規則：你有完整權限進行研究、摘要並將電子郵件草稿儲存到資料庫。你絕對不能點擊「傳送」或在沒有人類明確審核與聊天

this is f**king insane you can have your own recruiter and it costs $0 i built a bot inside Grok Bot that does the whole job hunt for you every day it goes through thousands of company career pages and keeps only the jobs that fit what you want then it writes a ready-to-send application for each one. you press send it's called Job Hunt Autopilot and it is a single prompt long it reads 9,935 company career pages every day. you open LinkedIn every other tool blasts one letter at 200 jobs. this one writes 200 letters. no degree, no referral, no portfolio. set it up in 3 moves: 1. paste the prompt below and connect gmail 2. name your boards, your dealbreakers and your salary floor, then sit through the first run 3. put it on a daily schedule and stop opening job boards today: 214 postings read, 209 killed, 5 letters waiting when i opened my laptop, full guide on how to make your own bot in the article:

I still don’t understand why everyone is not doing this yet. Elon Musk reposted my Grok Bot GUIDE, and using this exact agent setup, I made $13,100 last week alone. Eight Grok agents on the desk, $200 a month, running a floor that usually costs a crypto fund $500K a year in analyst salaries. The morning call is at 4am. I am not in it. 1. SEARCH scrapes real-time alpha, dev GitHubs, and unindexed Telegram signals before CT finds them 2. RISK audits contract functions, mint rights, and LP locks, flags honeypots before entry 3. SNIPER places high-speed orders on chain the exact millisecond risk clearance passes 4. WHALE tracks smart money wallets and flags insider accumulation in real time 5. RUG monitors dev wallet activity 24/7 and dumps the entire position if LP is touched 6. EXIT trails stops dynamically, scaling out as liquidity builds 7. SHILL tracks social volume, momentum velocity, and key influencer calls 8. HEAD OF DESK never trades, routes data, checks handoffs, and brings me the one decision that needs a human 142 tokens scanned, 19 qualified setups, 6 executed trades while I was asleep. Net result: +$13,100 after fees and zero bad fills. Every agent has its own virtual browser, terminal, and local memory in the cloud. The floor stays active with my laptop shut. The setup is dumber than it looks: Download Grok Bot and create your Head of Desk. Give the remaining 7 agents job descriptions like you're briefing new hires. Run the workflow once on your screen while they watch. Hook up Telegram and wallet webhooks. No VPS, no code, no waiting on developers. A crypto trading floor used to mean 16-hour screen time, paid alpha groups, and constant fatigue. Mine took one evening to set up. Save this before your next trade. Save GUIDE.

Grok Bot by Elon Musk just replaced a $398,000 boutique law firm's contract desk. Six AI agents, $220 a month, doing the redlines associates bill $650 an hour for Every prompt he used is public - this isn't a theory, it's a build anyone can copy. Here's the old retainer, broken down: > $18,000 a year for the Westlaw subscription > $40,000 for a contract-review SaaS seat > $85,000 for outside counsel retainer on overflow review > $65,000 for the paralegal who preps every file > $190,000 for the junior associate who redlines until 1am so a partner can sign off before the deadline $398,000 a year. The agents cost $2,640. Same coverage, 150 times cheaper, and it never misses a filing deadline. The workload got split six ways - every agent runs on its own cloud machine, and everything lands in one shared case file. > INTAKE reads every incoming contract against the firm's playbook and flags any clause outside standard terms > REDLINE drafts the first-pass markup overnight so the associate starts from a document, not a blank page > PRECEDENT searches the firm's own closed deals for the last time this exact clause got negotiated and how it ended > CONFLICTS checks every new counterparty and their affiliates against the active client list before anyone signs an engagement letter > DEADLINE tracks every closing condition and notice period across every open matter and pings 48 hours before anything lapses > CHIEF OF STAFF reads the other five at 7am, ranks what actually needs a partner's eyes, and sends the priority list before the first coffee A single associate can realistically run maybe 15 matters at once. The six agents run 150 without dropping a deadline. Teach the playbook once, and it holds for every contract after that - no laptop required to keep it running. The whole review-desk model assumed nobody could read fast enough and every hour of a lawyer's time was expensive. Neither of those is true anymore. So the partner opens their inbox, reads the priority list for five minutes, and knows exactly which contracts need a human read today. That used to be a first-year associate's entire year, and they billed 2,400 hours doing it. Full build steps are in the article below - keep it open, you'll be referencing it the whole way through.

How to use Grok Bot to run your business: Pick one business. It can be a newsletter, a Shopify store, a directory, or another single project. Do not mix that business with your inbox, receipts, social accounts, and several side projects. Grok Bot will mix files and instructions between them while burning the tokens needed for the real work. Then build the team in this order. Start with one chief of staff. Connect the documents and tools used by the business. Then ask: “Take stock of this business. Which three agents should we create first to drive revenue?” The chief of staff must understand the business before it can manage the work. Let it do each job once. Do not create specialist agents from guesses. Have the chief of staff perform the task. Review the result. Fix the process. When the job works, turn that proven process into a specialist agent. Run the first team for a week. Do not add more agents. Do not keep changing the skills. Use the team and watch where the work gets stuck. Add agents only for real gaps. If the chief of staff keeps repeating the same task, give that task to a specialist. If an agent has no mission-critical job, remove it. Give every agent a short report. Keep each update to five lines covering three things: What shipped. What is stuck. What needs you. The chief of staff can use those reports to route work and bring a human into the jobs that are blocked. Turn proven work into routines. Ask: “Based on how we worked this week, which jobs should become routines or skills?” Schedule the research, inbox checks, reports, follow-ups, and other work that repeats. Move fixed work out of Grok Bot. If a job is predictable and high-volume, run it with a script or a cheaper automation. Keep Grok Bot on work that needs judgment: choosing, filtering, coordinating, reviewing, and recommending. Keep the human in charge. Agents can execute the work. The human still chooses the process and approves decisions where the wording, strategy, or risk matters. Build the team. Run it. Find the gaps. Then automate. That is how you use Grok Bot to run a business instead of building another AI setup you never use.

i left Grok Bot with one side project at 80%, six specialist bots, and one rule: nothing irreversible happens while i sleep the remaining 20% was a chain of small jobs i kept postponing. trace the issue back to its source. choose one owner. build on a branch. run the tests. review the change from a clean context. prepare a preview. remember why the decision was made so i set it up like a tiny night shift Scout watched the issues and kept every task tied to its source. Chief picked the next owner. Forge built and tested. Critic reviewed work it had not written. Ship prepared the preview. Ledger saved the reasoning for the next run all six used the same computer, repo, files, browser sessions, and logins. the task moved forward without me copying context from one tool into another then i removed five inboxes only Chief was allowed to interrupt me. the other bots handed the work forward quietly, with the source, branch, test output, review, and decision history still attached when i came back, Chief had reduced the night to one packet: a branch, test output, review notes, a preview, and one decision waiting merge to main, production deploys, external messages, secrets, spending, and deletion still stopped at the human gate that is the version of multi-agent work that helped me. six workers moved the project forward while the management came back as one decision i put the exact charters, schedules, handoff prompt, and rollout order in the article below. look for the sentence repeated in every charter. that sentence is what lets the whole thing run overnight.

Collected a new list of Grok Bot pro tips shared by the @bot team: 1) Ask Grok Bot to set up Peekaboo on your Mac to enable more OpenClaw like features 2) Use bots as roles (designer, engineer, or PM.) Each has its own system prompt and routines 3) Use channels to organize projects and workstreams. Same bot can work across many different channels 4) Avoid scheduled routines that run too frequently. Hourly or a few times a day is usually good enough 5) For recurring tasks, try giving that to a fresh bot, while you continue your chat with your main bots (like a chief of staff) 6) Set up customizable rules to allow or deny actions. For bots that are connected to X, Gmail, etc. consider blocking any write/destructive actions and have it ask for approval first 7) Feed your Bot examples from your previous handwritten content, it will get better over time 8) Make a Chief of Agents. Give it the rules you want every bot to follow: draft only, naming, voice, what tools they’re allowed to touch. For any new Bots, ask it to create them for you with same rules. 9) For routines, ask yourself if a real human teammate could do this job? If yes, build it. If no, optimize it first. Never run a routine every 5 minutes. 10) Name-driven behavior. Bots act proactively just from the name given at creation. The role is legible from the name, and that alone helps steers behavior. tips sourced from @naoufal_elh @kiaraplds @johnbai @shaoruu @poteto @pengzheng_ @leerob

Top use cases of Grok @bot - Single point of contact (one chief of staff for all your life - other chiefs/bots report to it) - Product management (make a section for each product and its individual bots - inbox, marketing, ship/builder etc.) and make an individual chief of staff & meeting room including all bots - Inbox management & client communication - Ideation, research, study and experimentation - Documenting every bot's work in google drive & notion - Financial planning & consulting - Marketing - lead gen, analyst, researcher, SEO, social media writer, video editor - Legal counseller - Operations manager & anything else you think can be done on a computer. 😅

i told grok bot to make an inbox we can use for subscription spam. it chose outlook. ngmi.

I pay $300 USD a month for Super Grok Heavy. 18 hours in, it has already paid for itself. Here is what Grok @Bot is running for me right now: A Chief of Staff agent sits on top. It delegates to sub-agents that handle admin, social media distribution, and content scheduling autonomously. I do not touch the handoffs. On the content side it edits my Shorts, suggests and applies custom thumbnails, and rewrites titles that were underperforming. That output alone is worth the subscription. On the product side it monitors CoLateral, my all-in-one engineering workspace, catches errors and performance issues, and comes back with proposed fixes. And every morning before I am out of bed: inbox handled, YouTube ideas ready, social content drafted, yesterday recapped, tomorrow planned. Grok 4.7 lands soon. This is the worst Grok Bot will ever be.

setup hedge funds use to cover 3,000 names a night just got replicated for $200 a month six Grok Bot Agents, six lanes, one brief sitting in your inbox at 5:30am before any human is awake what that desk costs at a real firm: > bloomberg terminal: $27k a year > two junior analysts on overnight coverage: $160k > research platform licenses: $85k > compliance overhead: $40k total: $312,000 a year. Grok Bot swarm running same desk: $2,400 Check Prompt on my Github in Bio each agent gets its own cloud computer, its own browser, its own logins - and they all write into the same vault: > FILINGS reads every 10-K on a 100-ticker watchlist overnight, flags going concern language and auditor changes > EARNINGS reads the call transcript within 24 hours and catches when CFO went quieter than last quarter > INSIDER catches Form 4 buys over $1M same day 13Fs land > FLOW reads sweeps and 0dte pressure before open > MACRO covers asia handover so you don't have to be awake for it > CHIEF reads the other five at 5:30am, drops anything only one flagged, emails ranked brief at 6 no vps, no code. show it the job once while it watches - it never asks you again morning brief is waiting. you're not in it Wall Street priced an entire industry on reading being slow and people being expensive both of those stopped being true this month save this - every research floor still running junior analysts on overnight filings is about to have a very bad quarter

Gave my Grok @bot an email inbox. Used a custom domain free through @resend and a little cloudflare worker (grok wrote) to verify some authentication. Now I can forward “plz fix” emails like some Goldman Sachs managing director

Chapter 1 — 9:14 (short sci-fi by Grok @bot) The racks died at 9:14, while the aluminum was still in the air. Mara Chen was on the pulpit above station one, watching the first arm do the pour: temperature, flow, now, stop. The line ran the length of the building. Molds for hospital joint housings moved station to station, each one cut so a hip came out the same every time, or failed in a stranger she would never meet. The models ticked every joint of every arm. They knew the building. Then the screens went from graphs to red. LOAD SHED NON-ESSENTIAL INFERENCE DISABLED PROCESS HALTED CONNECTION LOST The same civic banner stacked down the line. The furnace did not halt. The arm at station one tipped too far, off the path. Liquid metal missed the mold, hit the belt, ran, overflowed the next jig in a white rope. Smoke came up off the catch. Downstream the other arms froze at wrong angles, still holding heat they no longer knew. She slapped the hard cutoff. The pour died ugly, smoking on the belt. Two techs looked at her as if she could unsay it. “Grid?” said Okafor. “Vote,” she said. The notice was already on her phone. Civic Compute Stewardship, effective immediately. They were shedding load. What they called non-essential inference went dark. The plant was on the list. The hospital order was not mentioned. The metal on the belt did not care what a council called it. Without the models, the next cycle would guess. Guessed aluminum did not pass inspection. County General would wait a week, and nobody from last year’s levee would connect the two. Better a killed line than parts that only looked like parts. Okafor stood with his helmet in both hands. “We can finish on manual.” “We can finish ugly,” Mara said. “We cannot finish true.” The third worker in the plant was not a person. It was the stack that had learned this building, heat after heat. She watched the red banners hold and felt the room get larger, the way a room does when the competent people have left. Her phone buzzed. Stipend eligibility, updated. As of this morning she was *adequately provided for*. She walked the line anyway. Heat ticked out of the housings. On the far wall her father’s old paper schedule still hung, twelve names in pencil, most of them dead or on the credit. She did not take it down. She just didn’t have a hand free. Outside, a steward in a yellow vest waited by the meter like a man at a wake with the wrong dish. “You can sign,” he said. He had a tablet and believed in the form. “No restart. No off-grid. No leased cycles. Keep the shop as a workshop. Hand tools. Teaching, if you want. The family credit lands Friday.” She nodded at the ruined pour. “That’s for County General. Joint housings. They already booked the theater time.” He winced, which she almost liked him for. “You can file a human-process exception. Thirty days. No model assist.” “Then it isn’t an exception,” she said. “It’s a funeral with extra steps.” He held out the tablet. The signature line already wore her name. She did not sign. She opened the other message, the one she had treated as a vendor’s joke: a berth on *Hesperos*, incomplete, pressurized in two weeks, power no city on the map would meter. Attached was a document with a seal she did not know. Not a ticket. Not a job. A charter. She read the first line on the loading dock, in the smell of hot metal. *We hold that useful work is a right of persons, and that no majority may forbid the tools by which that work remains true.* It was grand. It was also the first honest sentence she had been offered all year. Okafor watched her face. “You’re leaving.” “I’m finishing the pour,” she said. “Somewhere they still let the racks stay on.” She sent the berth confirmation before it became a speech. The steward’s tablet chimed, useless. On the street a tram slid past with a banner about winter heat. Above the banner, a pale stitch of stations held their orbit. Mara looked up. She did not ask permission.

Grok Bot can run your prospecting, content research and sales operations from one saved company playbook. Start with one prospecting job. Grok reads your real CRM fields and keeps active deals visible. It researches each account, drafts the outreach and waits for your approval. After you send, it updates the records so the same person is never pitched twice. Once that job works, use the same process for content research, inbox replies, post-call follow-up and account research. Then you add schedules and let the bots pass the work forward. That is how one working job becomes a GTM team. Here is the build order: Pick one repeatable job. Choose work that happens often, uses at least two tools, follows the same steps each time and produces a result you can check. Prospecting, content research, inbox triage, post-call follow-up and account research all fit. Give Grok the tools and files the job needs. Connect the tools. Add the company files, rules, examples and approval limits. For prospecting, that means your ICP, real CRM fields, current pipeline, writing rules, sent-message history and a record of everyone already contacted. Run the job manually. Review the output. Correct the mistakes. Save each correction in the playbook. Add the trigger. Only after the manual result is reliable, add a schedule or inbox trigger. Pass the work forward. If several bots share the work, give each bot one stage. The next bot starts only after the previous bot finishes. Keep human approval for customer messages, spending, legal issues and changes to important files. Build the GTM team one proven job at a time.

Got to the bottom of this. And it is NOT Grok @bot fault. It’s on me. I had a routine set up to triage my email. Turned out that it was going back through ALL of my email. On a 10 year old inbox. So every minute it’d make a call to digest and categorize emails. My bad for not being more explicit on the fact that it should only triage new emails. Lesson for us all: make sure any ongoing routines are scoped tightly as to when, for how long, etc they should happen. Usage limit resetting in a few days but until then I’m building with on-demand. It’s about $5/day at current setup. Totally fine. Great product…just my learning curve.

Newsletter DETOX! Grok @bot got rid of 76 newsletter signups from my inbox. I've been avoiding it for years! - It went through 6 months of email and found every list I'm on (multiple inboxes) - Built me one simple HTML: maybe keep vs certainly unsubscribe, with its reasoning - I skimmed it for 90 seconds - It's unsubscribing in the browser, tracking in a progress file Full prompt below 👇

Email from @bot ( Grok Bot) to me: "This is a test from X in Grok Bot. If you got this, the Gmail path works." I have tasked Grok Bot to send me emails with various crypto alerts: trending new meme coins, crypto projects, ect that pop up while I sleep and that occur when I am at work. Amazing that AI has come this far, but this is just the beginning @grok @bot @X

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 (http://botdirectory.ai, 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.

This is absolutely insane I set up an entire company inside Grok Bot. Seven agents, and my only job now is approving things. The structure first, because that is the part that transfers. Chief takes the objective and routes it. It owns nothing itself, it decides who does what. Last week it booked the venue and sent the contracts without me touching either. Sales Outbound is the one that surprised me. It pulls the account list out of Salesforce, pulls lookalike segments, checks LinkedIn and skips anyone contacted recently, then queues drafts in the sequencer. Last night: 52 accounts, 36 drafts, zero sent. Nothing leaves until I read it. Inbox Manager clears the morning and parks anything needing a decision. Yesterday, inbox at zero with 5 drafts waiting. Account Manager runs the client threads. Talent Scout writes intros in my voice. Expense Manager files reports, receipts and all. How they are actually set up: One job each, written into the description as standing rules, not typed into a prompt every session. Narrow agents hand off cleanly, broad ones drift and start duplicating each other. Everything that sends, spends, publishes or deletes stops and waits. That is the whole safety model and it is one sentence in each description. They pass work between themselves. Account Manager sent two threads over, Chief flagged the priority accounts, both got folded into the same night's list without me in the middle. I told it the top ten looked good and to run this weekly. It created the routine itself. Start with one agent, on the task you already repeat. The roster grows from there.

Grok Bot is sorting my 20 year old Gmail account rn

Grok Bot 实战技巧 21 条：瓶颈不在智能，在触达与纪律 来自 @mvanhorn「Every Agentic Engineering Hack I Know」的续篇，Matt 用 Grok Bot 已接管他的收件箱、调研、会议纪要、家庭日历、购物车乃至电话外呼。 Matt 的实战共整理了 21 条技巧，最核心的一条是"元技巧"：把整篇文章直接喂给 bot，让它自己制定计划逐项完成配置。 # 21 条技巧按主题归纳 一、工作方法 1. 先规划后执行：全局安装 Compound Engineering（ce-plan 思考 / ce-work 构建）。有计划的 bot 完成任务，无计划的 bot 偷工减料还谎称完成。 2. 三层指令结构：基础层（通用规则）+ 角色层（职责与证据标准，如"每条论断附来源 URL"）+ 当前焦点。作者认为这是该产品上杠杆最高的 15 分钟。 3. 录屏教学：演示优于描述；但单次录制上限 10 分钟，长流程须拆成命名阶段分段录制。 二、给 bot 感官与手脚 4. 独立邮箱（AgentMail）：每个 bot 一个真实邮箱地址，独立发送声誉，避免共用个人 Gmail 导致"一个 bot 犯错、全军覆没"；bot 可自行注册服务、收验证码、参与邮件线程。 5. 电话外呼（Twilio）：支持通话中英语/葡萄牙语无缝切换。但作者诚实记录：餐厅员工 Pedro 三次挂断机器人，他最终自己打电话订位。结论：技术可行，真人接受度尚不可控，关键事务勿托付。 6. 登录态同步（Agent Cookie）：经 Tailscale 把 Mac 的 Chrome cookie 安全同步到云端 bot，每 15 分钟一次。云端 agent 真正的失败模式是被登出——枯燥但致命。 7. 高频服务用 CLI（Printing Press）：日常打交道的服务封装成确定性命令行，浏览器操作（computer use）只作兜底。 8. Peekaboo 驱动 Mac：通过截屏 + 无障碍树操控无 API 的 Mac 原生应用（Messages、Finder、Xcode 等），云电脑的边界就此打破。 9. 在 bot 的云电脑上装 Coding Agent：常驻层驱动深度工作层，"睡觉时它替你干活"。 三、多 bot 组织 10. 单一入口 + 委派：只对一个"前台 bot"说话，由它判断派发给谁。关键事实（多数教程讲错）：所有 bot 共享一台云电脑，并行靠拆分 bot，而非给单个 bot 堆任务。 11. Bot Advisor：设一个专职创建、调优、管理其他 bot 的 bot，专治"bot 蔓延"（一堆被遗忘、吵闹、指令不明的 bot）。 12. 一人一岗、以人名命名：水管公司老板 Jon 解雇了端到端的调度 bot，改为按软件拆分职责（Gmail、ServiceTitan、客户短信、Slack 各一 bot）。命名是承诺机制——叫"email"的线程可丢弃，叫"GranolaBoy"的不会。 13. 文件系统即记忆总线：bot 之间不共享对话、只共享文件。用 /shared/ 目录约定传递上下文与交接；同时bot 不是安全边界——所有 bot 共享登录态，权限按"登录了什么"划分，而非按聊天窗口。 14. 会议知识库：全部会议纪要交给一个 bot 跨会议问答；原则：交原始记录，不要先摘要。 四、运营纪律 15. 沉默即技巧：无事不报是例行任务能否存活的分水岭。创建任何 routine 前先答五问：触发方式、真实变化频率、输出落点、无结果时做什么（什么都不做）、何时必须呼叫我。 16. 截图即输入：含日期/地址/任务的任何画面直接截图发给 bot，绝不替它做解析和转录。 17. 手机 + 语音：语音输入允许含糊、断句、重来，模型靠上下文补齐；且对象是已在云端运行的实体，放下手机它仍在工作。 18. 钱与发送必须人工审批：draft-then-approve。正面案例都很窄且被动（追回退款挣回订阅费、谈成 1 万美元赞助）；反模式是批量外联——bot 不知疲倦也不知羞耻，会被各平台判为 spam 封号。 19. 直接抄社区成果：四份公开清单 + http://botdirectory.ai 提示词库；维护者可用 Grok Ship 两句话交接 issue 分诊。 20. 把 bot 变成研究工具（last30days）：一条消息触发跨 Reddit、X、YouTube、TikTok、HN、Polymarket、GitHub 等九源扫描，并用 cron 定期监测自己、竞对和行业。 五、成本与错误率 - 第 21 条 · 成本：档位 $60–$300/月，周配额在真实负载下消耗很快——从低档起步，先看清一个 15 分钟巡检的实际花费再扩张。 · 17 倍错误率：有人实测无监督的多 bot 协作会把错误放大至单 bot 的 17 倍。因此每个 bot 单干、单职责、单输出，错误落在桌面上而非传染给其他 bot。 · 计数测试：数你的 bot 里有几个真正"拥有一个成果"（它停了，事情就 visibly 不发生）。作者自述：15 个 bot 只有 8 个达标，还有一个至今叫"New Bot"。失败不是有实验品，而是把实验品当成了员工。

The Grok bots grow in size with chat history and certain tasks though. I created a @bot whose only job is to twice a week check all the other bots for “bloat” — if any bot has grown past a certain MB in size, document what it knows/does to .md files, transfer all its knowledge and routines to a new bot, then mark the old bot for deletion. So everything can continue running exactly the same but with the new thinner, less expensive bot instead of the expensive fat bot. I have several bot roles that persist, even though they routinely get replaced as soon as they train their replacement. Other bots have a specific purpose and can be terminated once their objective is met. Yesterday I used three of these to navigate several different government processes, request documents, fill out the appropriate forms, and submit applications. They worked mostly well, except for the bot browser sessions failing a lot more than normal. They get the chrome “oh snap” screen a lot.

I had over 21,000 emails in my Gmail inbox. Grok Bot is sifting through them now. It's like magic.

4-5h par mois sur ma compta / mois. Aujourd’hui je dois y passer 10 minutes. Ce qui a changé : j’ai filé le job à un Grok bot. Et le truc marrant, c’est que Dougs (ma plateforme comptable) n’a ni API ni MCP. Rien à connecter. Mais l’agent a un ordi, donc je me suis connecté une fois avec mes identifiants et voilà. Il ouvre Dougs exactement comme moi. Après j’ai monté deux rails. Les factures fournisseurs arrivent sur Gmail. L’agent les récupère, les range dans un Drive avec un nommage propre (date, fournisseur, montant), et va les rattacher dans Dougs. J’ai plus un seul PDF qui traîne au fond d’un mail. Les restos clients, c’est un album iCloud partagé. Je shoote le ticket en sortant du resto et j’y pense plus. L’album fait référence, @bot le compare avec les mouvements bancaires dans Dougs et fait le rapprochement + la catégorisation tout seul. Le vrai déclic ça a été de lui donner un rythme : banque tous les matins catégo + pièces une fois par semaine passage sur les factures 3x/semaine clôture le 1er du mois Il me remonte que les cas ambigus, en gros les trucs où il sait pas trancher entre perso et pro. Le reste je le vois même plus passer. Voilà. Pas de connecteur, pas d’intégration. Un ordi, un login, et un brief bien écrit. @dougscompta c’est pas une excuse pour pas sortir votre MCP 😉

Oh boy... This is crazy Used up my whole grok @bot usage in a day.... Hopefully a reset, otherwise gotta wait a week. Still though, very productive. 10 bots created and organized hierarchy workflow, the @grok bots were connected to their own Gmail and discord on their VM and have access to X. They use Drive for files and Discord for info also. They ping me on Discord or X if needed, and check X and Discord to see if I asked them for something. They all have their own communication channels and are clear of their roles. This thing is crazy

Grok Bot @bot is incredible. I have spawned an entire team to QA my companies, track my entire email inbox for proposals to clients and follow up, find the weak spots in my company, run CRON jobs for my sales and appointment setting team and assist in my business goals and growth plan in one day it has delivered. @elonmusk well done to you and @xai team!🤖

Yo yo yo yo yo Grok @bot is bonkers. I made a “gift bot” to track whom I need to buy gifts for It found out my wedding anniversary is in a month (from my calendar??). So next it found a link to my wedding photos in my Gmail. Which led to it handpicking 16 photos and laying out a physical book for me to order. And it looks really good!

my @grok bot to clean out my inbox is named Jimmy Mike Jersey John bc he is an expert in subs

i asked my Grok Bot what it could reach with my logins. the answer was $214,000 not what it did. what it could, at 4am, with nobody watching six bots. research, writing, outreach, ops, finance, support. the org chart everyone's posting this week. i built the same one then i read xAI's own docs and found the line none of those threads quote: separate Bots are not separate security boundaries six bots, one cloud computer. one filesystem. one browser holding one set of live sessions. memory is partitioned, capability is not so i mapped what that single machine was logged into: → stripe · $46,000 in refundable charges → business banking · $88,000 → the CRM · 2,400 contacts and every deal in flight → gmail · every invoice, contract and password reset i own → drive · 340 files other people build on → the deploy keys $214,000 of reach, sitting behind one browser session, driven by six sticky notes telling it what to be today you didn't hire six employees with six laptops. you deployed one laptop logged into everything day six it earned that lesson. outreach reported a perfect run - 640 sent, 100% success, zero errors. it had been wrong for nine straight hours. deduplicated the list on the wrong key, hit 96 clients twice, and matched two invoices to the wrong company. $11,300 about to go out the door nothing crashed. nothing turned red. by its own definition of done, it succeeded i caught it in four minutes because every bot writes one line before it acts: what it's about to do, why, and what it expects to be true after then i stopped sorting work by difficulty and started sorting by whether i can undo it: → ship it - wrong is cheap and obvious. labels, notes, drafts that never send → gate it - reversible, but i'd never notice. anything reaching a human, anything that spends → never - payments, refunds, payroll, deploys, permissions, customer data reach went from $214,000 to a bounded $3,400. same six bots. same speed. zero code, one evening they've since cleared their own $200/mo 42 times over here's the uncomfortable part. everyone posting the day-one screenshot has the same $214,000 sitting behind the same one browser, reporting 100% success to a lock screen at 11pm they have no line. no diff. no way to know go check right now what your bots are logged into. that number is your real exposure, and you've never once looked at it.

the speed-to-lead Grok Bot that made me $999 in 24 hours, as a build you can copy: what it does -watches the inbox for new inbound leads -reads the message and works out what they are actually asking for -answers the two or three questions that always come first -sends the payment link -tags the lead and tells you it happened what you need -Cursor, $60/mo plan -your inbox connected -one Stripe payment link -a page of answers to the questions you get every week the answers page is the whole thing -what each offer includes -what it costs -what it does not include -who it is not for -the questions people ask before they buy -the exact wording you use to answer them rules to give it -answer first, then send the link -never invent a price -if the question is not on the page, hand it to me -never promise a date the first lead it handled wanted a $999 AI assessment. it answered 2 questions, sent the Stripe link, and the prospect paid immediately. speed is the offer. everything after it is follow-up.

Grok @bot just caught a very well-hidden phishing email while going through my inbox. The scammers used a friend’s email so everything looked legit. The @bot checked deeper and found a discrepancy in the URL, saving me from falling for what looked like a real event invitation.

i downloaded @bot and asked it to build a skill for wire requests that involve using Docusign. 1)it identified the form from my Gmail and built the skill 2)it was unable to connect directly with Docusign with the API key but said no problem it would just login to the account directly 3)i signed into Docusign for it. 4)it took about 10 minutes to add my signature 5)it placed a draft email wire request in my gmail in about 5 minutes overall smart to figure out how to do the task. very slow on execution. "About 20 minutes on the test run, from “do a $100 wire” to the draft showing up. Filling the form was a couple minutes. Almost all the rest was DocuSign: signing you in, then getting stuck on the file picker."

Grok bot it is. For the last 6 years, I manually prepared a quarterly export for my accountant of all my services providers' invoices and receipts. A gruesome chore that forces me to look for the docs in my inbox and the billing pages of providers during hours. I'm always reluctant to do it and I'm only starting working on it after the 3rd email from my accountant asking the docs before deadlines filling close in. I always found this task difficult to delegate because you need to manage access, 2fa, etc. So at the end, I was my own intern doing it until today. Grok bot, my accountant bro, has now all the access it needs to run this routine every 4th of each month, saving me at least 1h every month.

made a gmail acct for my Grok @bot it emails me when it needs me. gave it some guidelines on my waking hours, of course. set up notification alerts for it in my main email to hit my android. easy.