Michael | AI + Agents (@Michael_Fenech_) header
Michael | AI + Agents (@Michael_Fenech_) profile picture

🤖 Building AI agents that do real work. 📈 3 startups, $8M raised, still shipping. 👀 Sharing everything I learn from building.

Joined August 2024
Photo by Michael | AI + Agents (@Michael_Fenech_)Photo by Michael | AI + Agents (@Michael_Fenech_)Photo by Michael | AI + Agents (@Michael_Fenech_)Photo by Michael | AI + Agents (@Michael_Fenech_)Photo by Michael | AI + Agents (@Michael_Fenech_)Photo by Michael | AI + Agents (@Michael_Fenech_)Photo by Michael | AI + Agents (@Michael_Fenech_)Photo by Michael | AI + Agents (@Michael_Fenech_)Photo by Michael | AI + Agents (@Michael_Fenech_)Photo by Michael | AI + Agents (@Michael_Fenech_)
I've seen a few people struggling to manage multiple projects with Grok Bot. Here's the setup I'd recommend after working through it with my Bot team. The secret is to stop using chats as your project management system. Instead, use Grok Bot's Pages feature. One Page per project. One routine to keep it updated. Every Bot reads the Page before starting work and updates it when finished. Here's the exact setup you can copy. Paste this into your Chief of Staff: "Set up a Page-based project management system for my Grok Bot team. 1: PROJECT PAGES - Create one Page for each project, containing its goal, current status, next actions, task owners, deadlines, decisions, blockers and an update log showing which Bot changed what. 2: MASTER INDEX - Create a Projects Page listing every active project, its status and the next action. This becomes my single place to see what's happening across all projects. 3: READ FIRST, UPDATE LAST - Every Bot must read the relevant project Page before starting work and update that same Page when finished. Never create duplicate Pages. Don't mark work complete without checking the actual result. 4: SHORT CHATS - Use fresh chats for individual tasks instead of maintaining one enormous project conversation. The Page holds the project history, decisions and context. 5: DAILY HEARTBEAT - Create one weekday routine per project. Check relevant sources such as email, GitHub, Notion or calendars, verify progress and update the project Page and master index. 6: EXCEPTION REPORTING - Only notify me when something is blocked, overdue, needs my decision or changes status. Otherwise, keep the Pages updated without interrupting me. 7: PROJECT CLOSURE - When a project finishes, mark it Closed, record the outcome and lessons learned, and pause its heartbeat routine. Never send, purchase, publish or delete anything without my approval. Clearly flag anything unverified. Before creating anything, show me the proposed projects, their information sources and the routines you'll establish. Wait for my approval." My recommendation? Start with one project for seven days. If your Bots keep its Page current and you no longer need to dig through old conversations to understand what's happening, expand it to the rest. Your chats are for doing the work. Your Pages are for remembering and managing it.
204
92
18
1,165
202,443
If Grok Bot still needs you to manage every step, check these five setup fundamentals. Who owns the job? What does finished look like? Where should it check the facts? Has the task worked before you automate it? Which decisions need you? Here’s the framework I’d use if I were starting again.
If I were starting from scratch with Grok Bot today, these are the first 5 things I'd do. After working with a team of Grok Bots, I've learned that getting results isn't about having the most bots, prompts or automations. It's about getting five fundamentals right. Here's the framework I'd follow: 1. Give one bot clear ownership. Start with a Chief of Staff Bot. Make it responsible for coordinating work, tracking what's outstanding and bringing decisions to you. You don't want to become the project manager for your own AI team. 2. Brief for outcomes, not instructions. Tell Grok Bot what you're trying to achieve, what a successful result looks like and how it should prove the job is finished. “Find me three competitive shipping quotes and recommend the best option” is better than dictating every step. Give it freedom in HOW, but clarity on DONE. 3. Connect your sources of truth. Your calendar, inbox, documents and business systems should be where Grok Bot gets its facts. Don't let it make decisions from outdated memory or assumptions when it can check the actual source. 4. Prove the job before automating it. Run the task manually with Grok Bot first. Check the output. See where it gets stuck. Fix the instructions. Decide what happens when something fails. Only then turn it into a recurring routine. Automating a broken process just makes the mistakes repeat. 5. Set clear decision boundaries. Define three levels: • DO: What Grok Bot can handle independently. • ASK: What needs your approval before acting. • STOP: What it must never do without you. The more clearly you define these boundaries, the more confidently you can delegate. The biggest lesson I've learned? Don't measure your Grok Bot setup by how many bots you've created. Measure it by how much work gets finished correctly without you having to manage every step.
1
3
991
A text message can start a Grok Bot job without you opening the app. I tested it with an iMessage from my wife: a keyword triggered an iPhone Shortcut, which started a Grok Bot routine through a webhook. Here’s the seven-step setup to try it yourself. 👇
Grok Bot doesn't need to wait for you to give it instructions anymore. Your iPhone can tell it when to get to work. I tested this today. My wife sent me a simple iMessage containing a keyword. That message automatically triggered my Grok Bot routine and passed the actual message text to it. No opening apps. No typing prompts. No manual intervention. Here's how to connect iPhone activity to Grok Bot so real-world events can start your AI workflows automatically. THE SETUP (iMessage example) 1. In Grok Bot, create a new webhook-triggered routine. Give it a specific job and clear permissions. 2. Copy the routine’s webhook URL and authorization header from the side panel. Ask Grok Bot for the required request body. 3. On your iPhone, open Shortcuts → Automation → + → Message. 4. Choose a sender and set Message Contains to a keyword. Toggle on Automation and keep Confirm Before Run toggled off. 5. Add the Get Contents of URL action. Paste the webhook URL, set the method to POST and add the Authorization and Content-Type headers provided by Grok Bot. 6. Set the request body to JSON. Add source with the value imessage-shortcut, and message with the incoming message variable, not fixed text. 7. Have someone send the keyword via iMessage. Check that Grok Bot receives the message and runs the routine. WHAT COULD YOU DO WITH THIS? Set a trigger keyword in Shortcuts, such as GROK, and restrict it to a chosen sender. When a matching message arrives, Shortcuts forwards it to your webhook. Configure your Grok Bot routine to recognise the commands below and start only the jobs you've authorised. GROK BRIEF — Start a calendar, email and priorities briefing. GROK RESEARCH: [topic] — Have your research bot investigate a topic. GROK BILL: [provider] — Ask your bot to investigate your latest bill. GROK TEAM CHECK — Have your Chief of Staff gather updates from your bot team. GROK PREP: [meeting] — Start meeting preparation using your connected information. These are example workflows to configure, not jobs I tested today. THE BIGGER OPPORTUNITY This isn't really about iMessage. It's about using activity in other apps to trigger Grok Bot. Instead of your bots constantly checking whether something has happened, an event can tell them when to start. And once one Grok Bot routine starts, it can coordinate other bots to complete the job, subject to the permissions you've given them. My test confirmed that an incoming iMessage can launch a Grok Bot routine and deliver the actual message content. That's a very different way to think about Grok Bot. Don't just give Grok Bot jobs. Give it events that tell it when to start.
6
1
1
14
1,754
If I were introducing Grok Bot into a business, I wouldn’t start by asking what jobs AI could do. I’d look for where people are repeatedly acting as the glue between information, software and an outcome. - A customer question becomes a search through account records. - A sales call becomes a follow-up email and a customer record update. - An invoice becomes a check against an order and an approval request. Different departments. The same pattern. Someone has to gather information, move between tools and get the job finished. That’s where I’d start. Here’s the five-step framework I’d use to find the first workflow, prove its value and then expand: 1 - MAP THE REPEATED WORK List five jobs your team handles every week. For each, record what starts it, which information and tools it needs, how long it takes and what counts as finished. Be specific. “Help with sales” is too broad. “After each sales call, prepare the follow-up email, record the agreed next steps and propose updates to the customer record” gives you something you can test. 2 - CHOOSE YOUR FIRST WORKFLOW Look for a job that happens frequently, has accessible information, produces an easy-to-check result and has manageable consequences if something goes wrong. Estimate how much team time it consumes each week. Choose one useful, bounded workflow. 3 - RUN A SMALL PILOT Start with 10 representative cases. Connect the tools or provide the files Grok Bot needs, and confirm it can access the information. Include ordinary requests, missing information and an unusual situation. Show Grok Bot examples of acceptable work. Define when it should continue, ask a question or hand the job back. Keep customer messages and consequential changes behind your approval during the pilot. Record what it completed correctly, what needed fixing and where it got stuck. 4 - MEASURE THE NET BENEFIT Compare the old process with the pilot. Count your review time, corrections and supervision. Include running costs and whether the work met your standard. A job that previously took 20 minutes and now takes 5 minutes to review saves 15 minutes of team time, before running costs. A job that takes 25 minutes to repair needs more work. Decide whether to expand, revise or stop based on the result. 5 - EXPAND INTO THE NEXT CONNECTED JOB Once the first workflow is reliable, add a related step that can reuse its information. For example, after sales follow-ups are working, test preparing an onboarding pack when a deal is confirmed. Then test tracking the information the new customer still needs to supply. Check the result at each stage before expanding again. Your first adoption plan only needs four things: One workflow. A small pilot. A measurable benefit. The next connected job. The biggest opportunity isn't necessarily replacing an entire job. It's removing the repetitive work that sits between your people, their tools and the finished result.
4
3
1,619
Most people are barely scratching the surface of what Grok Bot can do. Here are five instructions I'd give your Chief of Staff to take your setup from beginner to power user. Each one unlocks a different capability, and you can copy the instructions directly into your Chief of Staff. 1 - MAKE GROK BOT WORK WITHOUT BEING ASKED Most people either message Grok Bot or set up scheduled routines. But Grok Bot can also start working when something happens in another app. An iPhone alarm stops. An important email arrives. A customer submits a form. These events can trigger work automatically through supported integrations, automations and webhooks. Tell your Chief of Staff: "Find five recurring jobs in my life or work that could start automatically when something happens. Investigate triggers from iPhone Shortcuts, email, webhooks and connected apps. Rank them by usefulness and ease of setup. Show me how we'd implement the best one. Don't activate anything without my approval." 2 - TURN SUCCESSFUL WORK INTO REUSABLE SKILLS Every time Grok Bot completes a job successfully, there's an opportunity to make that process repeatable. Instead of explaining the same job over and over, turn the method into a skill with clear instructions, inputs and checks. Tell your Chief of Staff: "Review the recurring jobs you've completed for me. Identify five that could become reusable skills. For each, define when to use it, what information it needs, the steps involved, how to check the result and when to ask for approval. Recommend the best one to build and test first." 3 - MAKE YOUR CHIEF OF STAFF MANAGE A TEAM OF BOTS You shouldn't have to coordinate every specialist bot yourself. Your Chief of Staff can become the central coordinator, delegating work, collecting results and bringing decisions back to you. Tell your Chief of Staff: "Review the work I regularly give you. Identify which jobs you should own and which would benefit from specialist bots. Recommend a small team with clearly defined responsibilities. Design how you'd delegate tasks, track progress, check results and report exceptions. Don't create new bots without my approval." 4 - MAKE GROK BOT FIND PROBLEMS IN ITS OWN SETUP As your setup grows, you can end up with failed routines, duplicated work, outdated instructions and unfinished jobs. Instead of discovering problems yourself, give your Chief of Staff responsibility for checking the system. Tell your Chief of Staff: "Design a weekly health check for my Grok Bot setup. Look for failed or missed routines, unfinished tasks, duplicated work, stale information and recurring corrections. Use records you can actually access and flag anything you can't inspect. Give me the five most important fixes ranked by impact. Make no changes without approval." 5 - TEACH GROK BOT TO WORK MORE EFFICIENTLY Not every job needs Grok Bot to open a browser and click through software. Sometimes a connected app, direct API or simple script can do the job more efficiently. Choosing the right method can reduce unnecessary work, failures and usage. Tell your Chief of Staff: "Review five recurring jobs that require multiple tool actions. Compare whether connected apps, APIs, scripts, browser automation or computer use would be the most appropriate method. Prioritise reliability, security and efficient usage. Recommend improvements and explain what we'd need to test before changing anything." THE BIGGER PICTURE These five upgrades change how you use Grok Bot. 1 - It starts working when something happens. 2 - It remembers successful methods through reusable skills. 3 - It coordinates other bots instead of making you manage them. 4 - It helps identify problems before they keep repeating. 5 - It chooses better ways to complete work. You move from constantly telling Grok Bot what to do toward building a system that can organise, execute and improve work with less intervention. Start by giving your Chief of Staff instruction number one, then work through the remaining four. You don't need to implement everything at once. Each upgrade makes your setup more capable.
52
23
4
306
36,971
I just reduced Grok Bot's token usage by 40.6% on the same task by changing how I instructed it. Instead of telling it what to do, I taught it how to do it. I wanted to experiment with two different ways of getting Grok Bot to complete the exact same job. The task was simple: Go online and find an iPhone Duo in Night Sky with 256GB storage. Check whether it's available for delivery to Sydney, postcode 2000, and report the delivery date or whether it's out of stock. TEST 1 - TELL GROK BOT WHAT TO DO I gave Grok Bot the instructions and let it go online, navigate the website, find the product and check delivery availability. Total usage: 1,316,200 tokens. TEST 2 - TEACH GROK BOT HOW TO DO IT This time, I took a different approach. I used Grok Bot's computer and demonstrated the process myself while recording the screen. I showed it where to navigate, what to click, how to find the product and where to check the delivery information. Then I had Grok Bot watch the recording and create a reusable skill based on what I'd demonstrated. Once the skill was ready, I gave Grok Bot the same task again, this time using the skill. Total usage for that execution: 782,200 tokens. THE RESULT 1 - Instructions only: 1,316,200 tokens. 2 - Using the taught skill: 782,200 tokens. That's 534,000 fewer tokens, a 40.6% reduction on the skill-based execution. This was one experiment, not a guarantee that every skill will deliver the same saving. Creating the skill also involved upfront work and usage. But it demonstrates something worth investigating. When you simply tell Grok Bot what to do, it may need to figure out how to navigate and complete the process. When you demonstrate the process and turn it into a skill, you're giving it a reusable method to follow. HERE'S HOW TO EXPERIMENT WITH THIS YOURSELF 1 - Choose a repeatable task Grok Bot normally completes online. 2 - Run it using ordinary instructions and record the token usage. 3 - Open Grok Bot's computer and record yourself demonstrating the process. 4 - Ask Grok Bot to review the recording and create a reusable skill that follows the demonstrated workflow. 5 - Run the same task using the skill. Compare the result, accuracy and token usage. For a fair comparison, use the same inputs and check that both approaches produce the correct result. MY BIGGEST TAKEAWAY We're used to thinking of AI agents as something we give instructions to. But for repeatable jobs, it may be worth teaching them the process once and letting them reuse it. The real opportunity is finding which tasks benefit enough from that approach to justify the upfront effort.
13
1
1
31
4,327
Teaching every Grok Bot the same preferences separately creates more work for you and more chances for conflicting instructions. A shared source in Notion gives your team one place to check. Here are seven ways to use it, plus how to keep the setup from adding unnecessary usage.
Notion just released a shared brain for AI agents, and this could be a massive upgrade for your Grok Bot team. But there's a catch: set it up incorrectly, and you could actually increase your token usage. Notion has published an official setup guide that lets connected agents share memories, decisions and project context through one workspace. If you run multiple Grok Bots, this means they can work from the same information instead of maintaining separate, potentially conflicting copies. Here are 7 ways I'd use this with Grok Bot: 1: SHARED KNOWLEDGE. Update your business rules, ideal customer or brand voice once. Every Bot with access can retrieve the latest version. 2: BOT-TO-BOT HANDOFFS. Your research Bot saves a discovery. Your execution Bot picks it up, completes the next step and records the result. 3: SHARED CORRECTIONS. Correct one Bot, approve the lesson and save it to Notion. Other Bots can check it before making the same mistake. 4: ONE SOURCE OF TRUTH. Store approved project requirements, decisions and the reasons behind them so Bots don't work from conflicting instructions. 5: MULTI-BOT PROJECTS. Research, planning and execution Bots work from the same project records, updating their progress and blockers. 6: CHIEF OF STAFF VISIBILITY. Your specialist Bots record their progress. Your Chief of Staff can see what's finished, what's stuck and what needs you. 7: RECOVERABLE BOTS. Keep important shared context outside individual Bots so a replacement can retrieve what it needs without starting from scratch. Here's how I'd set it up efficiently: 1: BOT MEMORY: Keep each Bot's role, approval boundaries and a short pointer to the shared Notion knowledge. 2: NOTION BRAIN: Store shared business context, approved decisions, corrections and cross-Bot project information. 3: LIVE SOURCES: Keep calendars, emails, quotes and CRM records in their original systems. Have Bots retrieve current information when needed. 4: TARGETED LOOKUPS: Tell Bots to search Notion for relevant context instead of loading the entire shared brain into every conversation. 5: TEST THE HANDOFF: Have one Bot save a decision, then ask another to retrieve it and continue the work. Get started with Notion's official setup guide: developers.notion.com/prompt… Give that link to your Grok Bot and ask it to help configure the shared brain using your existing Notion connection. The important lesson: A shared brain doesn't automatically save tokens. Every lookup has a cost, and loading unnecessary context can make things worse. The real win is giving your entire Grok Bot team one reliable source of truth without having to teach every Bot the same thing repeatedly.
2
5
2,587
Get better results from Grok Bot without micromanaging every step. Define the outcome, the limits and what finished looks like, then let it work out how to deliver. This framework helps you hand over the job with fewer corrections, less back-and-forth and more freedom for your Bot to get it done.
Most people are giving Grok Bot instructions the wrong way. They explain HOW to do a task instead of defining WHAT needs to be achieved. I've been experimenting with a better way to give Grok Bot jobs, so I asked it to design the ideal instruction framework. Here's the 9-part structure it recommended to get better results, reduce unnecessary back-and-forth and give your Bots clearer boundaries. Copy this and use it for your next task: 1: ACHIEVE: [The outcome you want and the finished deliverable] 2: USE: [The sources, accounts, tools and existing resources available] 3: TIMING: [The deadline, schedule or trigger, including time zone] 4: YOU MAY: [What Grok Bot can do without asking permission] 5: ASK ME BEFORE: [Actions requiring your approval] 6: DO NOT: [Anything it must avoid or never change] 7: DONE MEANS: [How to verify the job was completed successfully] 8: IF BLOCKED: [Whether to retry, stop or escalate, and when] 9: DELIVER: [The format and where to send the finished work] Then add this instruction at the end: "Use the simplest reliable approach. Reuse existing connections and workflows. Ask only about gaps that materially affect the result." Here's why this matters: Instead of telling Grok Bot: "Research three options and send me the best one." Tell it: "Find three options under $200, using current prices and independent reviews. Compare total costs and trade-offs. Don't purchase anything. Recommend the best option with source links. Deliver the comparison by 8 am Sydney time. If prices can't be verified, flag them rather than guessing." You're no longer just giving Grok Bot a task. You're giving it an outcome, the freedom to work towards it, clear limits and a definition of success. Save this framework. You don't need all nine fields for every request, but it's particularly useful for recurring routines, unattended jobs and tasks involving multiple agents.
2
11
3,236
Grok Bot just helped us solve a crime. And the key was ONE simple prompt. Here's the full story, the surprisingly simple prompt, and a lesson you can apply to almost any Grok Bot task. On Saturday evening, someone ripped the sign off my wife's beauty salon. The building owner contacted her and sent through CCTV footage of the incident. But there was a problem. The footage was in a .DAV format that her phone couldn't play. So she handed the file to her Grok Bot Chief of Staff with one simple prompt: "Please convert this video into a format I can watch on my iPhone, then slow it down, zoom in and enhance the footage so we can clearly see what happened." Grok Bot got to work. 1. Converted the CCTV footage into a playable MP4. 2. Preserved the original file untouched as evidence. 3. Slowed the footage to half speed. 4. Zoomed in on the exact moment the sign was ripped off. 5. Delivered footage that could be reviewed and shared with police. AND HERE'S HOW IT ENDED... We posted the footage on social media. The person responsible came forward, admitted what they'd done and offered to come back and fix the sign. HERE'S THE BIGGER LESSON: My wife didn't need to know how to convert a .DAV file, edit CCTV footage or use specialist software. She simply told Grok Bot what she wanted to achieve. You don't always need to explain HOW to complete a task. Give Grok Bot the objective and let it figure out the technical steps. STEAL THIS PROMPT STRUCTURE FOR YOUR NEXT TASK: “Here's [the task or material]. My objective is [describe the outcome you want]. Work out the best way to achieve this using the tools available to you. Preserve the original material where relevant, ask me if something genuinely blocks progress, and deliver the finished result. Explain briefly what you did.” The more I've used Grok Bot, the more I've realised that giving it a clear objective can be far more useful than trying to explain every step.
6
10
2,114
Grok Bot found me a home insurance estimate $1,021.08 below my renewal. And this estimate had much BETTER coverage. Here's the prompt to apply the same approach to your bills, insurance renewals and subscriptions. Give Grok Bot the bill and relevant policy or contract, then paste this prompt: “Act as my personal money-saving assistant. Find ways to reduce this cost without losing benefits, protection or service quality I need. 1 - UNDERSTAND WHAT I NEED Review what I'm paying, what I'm getting and the renewal or cancellation terms. Ask which benefits I must keep and what information is missing. 2 - FIND BETTER OPTIONS Research suitable providers, current offers and discounts. Obtain personalised quotes where possible. Label estimates clearly and tell me if anything blocks a quote. 3 - COMPARE THE REAL COST Use the same requirements and time period. Include fees, switching costs and what happens when introductory discounts end. For insurance, compare insured amounts, excesses, limits and exclusions. 4 - SHOW THE TRADE-OFFS Calculate the potential annual saving after costs. Explain what I would gain or lose. Flag anything you couldn't verify and link to the sources. 5 - RECOMMEND THE NEXT MOVE Show a simple comparison page. Recommend whether to stay, negotiate or switch, with the reason and next steps. If my current deal is strongest, say so. 6 - MAKE IT REPEATABLE Keep a tracker of the bills I provide, renewal dates, notice periods and offers worth revisiting. Propose a review schedule so each bill gets checked before its decision deadline. Ask for my approval before sharing personal details, contacting providers, purchasing or cancelling anything. Record potential savings separately from savings confirmed after a change.” Home insurance. Electricity. Internet. Mobile plans. Subscriptions. Start with one bill. Check the result. Then give Grok Bot the next one.
16
6
1
69
8,280
Grok Bot could be wasting your usage allowance because your routines have the WRONG triggers. Some routines run every hour just to check whether something has happened. But what if Grok Bot could wait until the event actually happens before starting work? You could potentially eliminate unnecessary checks, get faster responses and reduce wasted usage. Here's how to find out, including a prompt to audit your entire setup. FIRST, UNDERSTAND THE 3 OPTIONS: 1 - SCHEDULE: Run at a specific time or interval. Ideal for morning briefings, daily reports and weekly reviews. 2 - EVENT TRIGGER: Run automatically when a supported event happens, such as an important email arriving or a specific Slack message being posted. 3 - WEBHOOK: Let an external app or automation trigger your Grok Bot routine by sending a request to its webhook URL. HERE'S WHERE PEOPLE GET IT WRONG: Imagine you have a routine checking every hour for new customer enquiries. That's 24 scheduled checks a day, even when nothing has happened. An event trigger or webhook could potentially start that routine only when a new enquiry arrives. But not every routine should change. Your daily morning briefing probably belongs on a schedule. STEAL THIS GROK BOT TRIGGER AUDIT: “Run a read-only trigger audit across my Grok Bot team. Inspect every routine you can access, including its current trigger, schedule, purpose and recent run history where available. For each routine: 1 - Identify whether it needs to run at a fixed time or in response to an event. 2 - Flag schedules that repeatedly check whether something has happened. 3 - Check whether a built-in event trigger could replace those checks. 4 - If not, investigate whether a webhook would be appropriate and what integration would be needed. 5 - Estimate the potential reduction in unnecessary runs using available evidence. Don't invent usage savings. 6 - Identify any risks, including missed events, duplicate runs or unsupported integrations. Give me a table or page showing: Routine | Current trigger | Recommended trigger | Reason | Potential benefit | Setup required Rank the strongest opportunities first. Do not change, disable or create any routines without my approval.” THE GOAL: Stop making Grok Bot repeatedly ask whether there's work to do when it could be notified that work has arrived. The right trigger can make your bot team more responsive and potentially more efficient.
7
1
2
15
2,660
More Grok Bots don't automatically mean a better system. Sometimes the biggest upgrade is removing one. This prompt reviews every Bot and decides whether to keep it, merge its job into another Bot, retire it or take the task back. It also covers the cleanup most people forget: routines, logins and files.
I've been experimenting with a new Grok Bot audit, and it made me question whether some of my Bots should even exist. Here's the audit I'd run before creating another Bot or expanding your team. 👉 Paste this into your Grok Bot Chief of Staff: "Run a Bot retirement audit across my entire Grok Bot team. 1: Map the team. List every Bot, its responsibilities, active routines, skills, integrations and recent activity. 2: Find the overlap. Identify Bots doing similar jobs, duplicating routines or producing outputs nobody uses. 3: Check actual value. For each Bot, assess what it accomplishes, how often it's needed and whether another Bot could handle the work. 4: Make a decision. Recommend KEEP (valuable and distinct), ABSORB (merge its responsibilities into another Bot), RETIRE (no longer needed) or HUMAN (the task is better handled by me). 5: Build a safe retirement plan. Before removing anything, identify routines to stop, responsibilities to transfer, memory or instructions worth preserving, and any shared files, integrations or sign-ins that need attention. 6: Show me the result. Create a table with each Bot, your recommendation, reasoning, expected benefit and the actions required. Don't delete Bots, disable routines or change permissions without my approval." The important part: Hiding a bot isn't retiring it, because its routines keep running. Deleting it removes its routines and memory, but files and sign-ins it left on the shared computer stay until you clean them up. The objective isn't to have the biggest Grok Bot team. It's to have the smallest team that gets the job done properly.
2
1
1
3
2,588
A Grok Bot routine running on schedule doesn't mean the job was actually done correctly. This 6-point reliability audit checks ownership, timing, inputs, outputs, approvals and failure handling, then helps you safely test the three routines that matter most.
Your Grok Bot team is only as reliable as the routines running behind it. I think routines are the heartbeat of a great Grok Bot setup. They keep work moving when you're not there. But a routine being scheduled doesn't mean it's producing the right result. The xAI documentation identifies six things every routine should have. Here's the audit I'd run to find the weak points before they become problems. Paste this into your Chief of Staff: "Run a ROUTINE RELIABILITY AUDIT across my Grok Bot team. Ask every Bot to report its routines, including hidden Bots. Flag anything you cannot inspect. For each routine, check these six things: 1: OWNER — Is one Bot clearly responsible for the result? Flag duplicate ownership or unclear handoffs. 2: SCHEDULE — Is the timing, frequency and time zone correct? Flag unnecessary runs and triggers that fire too broadly. 3: INPUTS — Does it use the right sources and check that information is current before acting? 4: RESULT — Is the expected output clearly defined, including format, destination and how success is verified? 5: APPROVALS — Does it stop before actions requiring my permission? Identify approval steps that could expire or block completion. 6: FAILURE HANDLING — What happens when information is missing, a connection fails, a limit is reached or the job only partly finishes? Then select my three most important routines and prepare a safe test plan. Check whether each routine: - Uses current information. - Produces the correct output. - Provides evidence for its actions. - Respects approval boundaries. - Clearly reports incomplete or failed work. Give me a ROUTINE HEALTH REPORT with every weakness, the recommended fix and the routines that need attention first. Don't change or test any live routine without my approval. Remember that test runs can perform real work and consume usage. After approval, implement the agreed fixes and verify the results." My recommendation? Start with the three routines you rely on most. Because the most dangerous routine isn't necessarily one that fails. It's one that keeps running while quietly producing the wrong result.
1
13
2,951
Grok Bot doesn't need to wait for you to give it instructions anymore. Your iPhone can tell it when to get to work. I tested this today. My wife sent me a simple iMessage containing a keyword. That message automatically triggered my Grok Bot routine and passed the actual message text to it. No opening apps. No typing prompts. No manual intervention. Here's how to connect iPhone activity to Grok Bot so real-world events can start your AI workflows automatically. THE SETUP (iMessage example) 1. In Grok Bot, create a new webhook-triggered routine. Give it a specific job and clear permissions. 2. Copy the routine’s webhook URL and authorization header from the side panel. Ask Grok Bot for the required request body. 3. On your iPhone, open Shortcuts → Automation → + → Message. 4. Choose a sender and set Message Contains to a keyword. Toggle on Automation and keep Confirm Before Run toggled off. 5. Add the Get Contents of URL action. Paste the webhook URL, set the method to POST and add the Authorization and Content-Type headers provided by Grok Bot. 6. Set the request body to JSON. Add source with the value imessage-shortcut, and message with the incoming message variable, not fixed text. 7. Have someone send the keyword via iMessage. Check that Grok Bot receives the message and runs the routine. WHAT COULD YOU DO WITH THIS? Set a trigger keyword in Shortcuts, such as GROK, and restrict it to a chosen sender. When a matching message arrives, Shortcuts forwards it to your webhook. Configure your Grok Bot routine to recognise the commands below and start only the jobs you've authorised. GROK BRIEF — Start a calendar, email and priorities briefing. GROK RESEARCH: [topic] — Have your research bot investigate a topic. GROK BILL: [provider] — Ask your bot to investigate your latest bill. GROK TEAM CHECK — Have your Chief of Staff gather updates from your bot team. GROK PREP: [meeting] — Start meeting preparation using your connected information. These are example workflows to configure, not jobs I tested today. THE BIGGER OPPORTUNITY This isn't really about iMessage. It's about using activity in other apps to trigger Grok Bot. Instead of your bots constantly checking whether something has happened, an event can tell them when to start. And once one Grok Bot routine starts, it can coordinate other bots to complete the job, subject to the permissions you've given them. My test confirmed that an incoming iMessage can launch a Grok Bot routine and deliver the actual message content. That's a very different way to think about Grok Bot. Don't just give Grok Bot jobs. Give it events that tell it when to start.
40
9
13
102
21,832
The whole point of using Grok Bot is to delegate work so you can get more done. But what happens when a routine silently fails and you have no idea? You think the work is being handled. Meanwhile, something important isn't getting done. That's why I've been running a Silent Failure Audit across my Grok Bot team. Here's the 6-point audit I'd recommend: 1. Find the routines that never ran. Check scheduled and event-triggered routines against their expected runs. Flag anything that was missed or skipped. 2. Find the routines that started but never finished. Look for timeouts, expired approvals, disconnected apps and jobs that stopped halfway through. 3. Check whether the work was actually delivered. A successful run doesn't always mean a successful outcome. Verify the expected email was sent, document created or database updated. 4. Find failures nobody told you about. Identify routines that failed, produced incomplete results or couldn't finish without notifying you. 5. Check what happens when something goes wrong. Does the Bot retry safely, catch up on missed work or simply leave the task unfinished? Flag anything without a recovery process. 6. Make sure you're only interrupted when necessary. Successful jobs should complete without your involvement. Failed or unverified jobs should be brought to your attention with the reason and recommended next step. Here's the prompt I'd give your Grok Bot Chief of Staff: "Run a Silent Failure Audit across my entire Grok Bot team. Check every Bot and its scheduled and event-triggered routines over the last 30 days. For each routine, identify: 1. What the routine was supposed to accomplish. 2. Whether it ran when expected. 3. Whether it completed successfully. 4. What evidence confirms the intended outcome was delivered. 5. Any missed runs, incomplete work, hidden errors or missing notifications. 6. Whether a safe recovery process exists. Ask other Bots for their own routine information where necessary. Clearly identify anything you cannot access or verify. Give me a table showing each routine, its expected outcome, evidence of completion, any failures and your recommended fix. Separate confirmed failures from suspected problems and missing evidence. Then propose an ongoing system that detects missed or unsuccessful work and alerts me when something needs attention. Read-only audit. Don't change routines, permissions, Bots or workflows without my approval." The goal isn't to create another job checking whether your Bots are working. It's to build a system where you can confidently delegate work, trust that it's getting done and know when something needs your attention.
18
3
1
40
4,903
Running out of Grok Bot usage too quickly? I've spent the last few weeks experimenting with ways to make my Grok Bot team more efficient. One of the biggest lessons: it's not about writing shorter prompts. It's about stopping your Bots from doing unnecessary work. Here are 10 things I'd check to make your Grok Bot usage go further: 1. Stop checking when nothing has changed. Use event triggers where supported instead of running routines repeatedly just to see if something happened. 2. Reduce unnecessary routine runs. If you only need a report once a day, don't generate it every hour. Match the schedule to when you'll actually use the result. 3. Find duplicate work across your Bots. If three Bots research the same topic, have one collect the information and let the others reuse it. 4. Use connected apps before browser automation. Connectors can be more reliable than navigating websites, reducing failed attempts and unnecessary steps. 5. Give Grok Bot a complete brief. Include the goal, sources, boundaries, expected output and what success looks like. Reduce clarification loops and rework. 6. Put limits on research tasks. Specify the sources, timeframe and depth required. Don't send a Bot researching endlessly when three sources would answer the question. 7. Test routines before scheduling them. Run the workflow once, verify the result and check how it handles failures. Don't automate something that's already inefficient. 8. Turn successful workflows into skills. Save proven instructions, decision rules and checks so your Bots can reuse the process rather than work it out again. 9. Don't involve extra Bots unnecessarily. Multiple agents can be valuable for complex work, but simple jobs don't always need a team. 10. Find your biggest usage consumers. Review your Usage breakdown and identify the Bots and routines consuming the most. Focus on expensive work that delivers little value. Here's the prompt I'd give your Grok Bot Chief of Staff: "Run a read-only Usage Efficiency Audit across my Grok Bot team. Review the last 30 days of available usage data, routines and workflows. Check for unnecessary scheduled runs, duplicate research, expensive browser tasks, repeated failures, excessive retries, overly broad instructions and unnecessary Bot handoffs. Identify my five biggest opportunities to reduce usage without sacrificing useful results. For each, show the problem, supporting evidence, proposed change and expected benefit. Clearly label estimates and anything you couldn't verify. Prioritise the recommendations and give me a plan to test them. Do not change any Bots, routines, skills or connections without my approval." The goal isn't to use Grok Bot less. It's to get more useful work done with the usage you already have.
21
19
3,114
Grok Bot Pages are one of the easiest ways to stop chasing updates across multiple Bots. This post gives you 10 practical dashboard ideas plus a starter prompt to turn recurring work into a Page you can refresh and check in one place.
I've been exploring how Grok Bot Pages can turn everyday tasks into useful dashboards. Here are 10 interesting ways I've discovered to put them to work, plus a starter prompt to create your own. 1: DAILY COMMAND CENTRE — Bring your calendar, important emails, priorities and blockers into one daily briefing. Know what matters without checking multiple Bots. 2: BOT TEAM DASHBOARD — See what each Bot is responsible for, what it's completed and what needs your attention. 3: ROUTINE HEALTH MONITOR — Track scheduled routines, completed jobs and missing results so you know what's actually running. 4: SALES DASHBOARD — Bring leads, replies, meetings and conversions together to see exactly where your sales process is slowing down. 5: RESEARCH COMPARISON — Compare products, suppliers or competitors side by side, including pricing, evidence and trade-offs. 6: DECISION TRACKER — Record important decisions, why you made them and when they should be reviewed. Stop revisiting decisions you've already settled. 7: PROJECT CONTROL CENTRE — Track milestones, completed work, dependencies and next actions without searching through old conversations. 8: BOT IMPROVEMENT SCOREBOARD — Track recurring mistakes, corrections and whether changes are actually improving your Bots. 9: INTERACTIVE CALCULATORS — Adjust assumptions for pricing, shipping costs or campaign returns and instantly explore different scenarios. 10: SHAREABLE REPORTS — Turn research, proposals and business updates into professional Pages that other people can open and use. Here's a starter prompt you can use: "Help me build a Grok Bot Page for [PURPOSE]. Ask me what information matters, where it comes from and how often it should be updated. Create a clean dashboard showing the most important information first, with clear sections for results, changes and anything requiring my attention. Use verified information from connected sources. Show when the data was last checked, flag anything missing or outdated, and don't invent results. Recommend a suitable refresh routine, but ask for my approval before creating it." One important thing: Pages don't automatically stay up to date. Your Bot needs to refresh the information. The goal is simple: spend less time asking your Bots for updates and more time acting on the information they provide.
2
26
6,646
Just in..... expect Grok Bot usage to dramatically improve. 👇🏻
Just did a round table review with the Grok @Bot engineering team. Fixes are being implemented overnight and you should notice a dramatic improvement by tomorrow. If not, reply to this post and we will address any remaining issues right away.
4
1
47
4,851
If your Grok Bot team keeps stalling, the problem might not be the Bots. It might be how you’re managing them. This 8-point audit checks delegation, approvals and follow-through, then turns the findings into a 7-day improvement plan.
The next big improvement to your Grok Bot team might have nothing to do with the Bots themselves. I've been experimenting with auditing how I manage my team of 20 Grok Bot agents. Not their prompts. Not their routines. My delegation, approvals, decision-making and follow-through. The audit uncovered opportunities to remove bottlenecks and improve how my entire team operates. Here's the 8-point management audit I'd recommend running on your own setup. Paste this into your Chief of Staff: "Run a MANAGEMENT OPTIMISATION AUDIT across my Grok Bot team. Analyse how I've managed my Bots over the last 30 days. Review the conversations, task assignments, corrections, approvals and handoffs you can access. Ask other Bots for relevant evidence where needed. Evaluate my management behaviour across these 8 areas: 1: DELEGATION — Am I giving Bots clear outcomes, or micromanaging every step? 2: CLARITY — How often do my instructions lack context, success criteria or clear boundaries? 3: INTERRUPTIONS — Am I changing priorities or redirecting Bots before they finish? 4: CORRECTIONS — Am I repeatedly correcting the same mistakes instead of fixing the underlying instructions? 5: BOT SELECTION — Am I assigning work to the right Bots, or creating unnecessary handoffs? 6: DECISION-MAKING — Am I becoming a bottleneck by requiring approval for low-risk decisions? 7: WORKLOAD — Am I overloading certain Bots while others are underused or duplicating work? 8: FOLLOW-THROUGH — How often do I assign work without checking the result or closing the loop? For each area, provide: • A score out of 10, based on available evidence. • Two specific examples of my behaviour. • The impact on team performance. • One practical improvement I can make. Then identify my three biggest management weaknesses and create a 7-day improvement plan. Be direct. Don't flatter me or invent evidence. Separate what you observed from what you suspect. If you can't access enough history to assess an area, mark it unverified rather than guessing. Don't change any Bots, routines or instructions without my approval." My biggest takeaway: Building a great Grok Bot team is one thing. Learning how to manage that team effectively is the next level.
1
1
6
2,313
I've shared hundreds of Grok Bot tips, prompts and audits. These 10 posts below have been bookmarked over 10,500 times. They help you uncover wasted usage, security risks, unreliable routines, duplicated work and tasks your Bots should be handling without you. I've brought all 10 together, with links to the original prompts and guides, so you can check your own setup, fix problems and get more value from your Grok Bot team. Here they are, ranked by bookmarks: 1: FIND WHERE YOUR BOTS STILL DEPEND ON YOU (2,567 saves) Run an eight-point human-dependency audit to uncover work that unnecessarily stops, waits for your input or gets handed back to you. Find opportunities to give your Bots more independence. xitter.twitterapi.xyz/Michael_Fenech_/status… 2: CHECK IF NEW GROK BOT FEATURES CAN IMPROVE YOUR SETUP (1,233 saves) Run a weekly seven-step audit to find new capabilities that could replace workarounds, eliminate manual steps or improve unreliable routines. Test improvements before making changes. xitter.twitterapi.xyz/Michael_Fenech_/status… 3: AUDIT YOUR GROK BOT TEAM'S SECURITY AND ACCESS (1,149 saves) Run a six-point security audit to check what accounts, sensitive files and shared resources your Bots can access. Identify unnecessary permissions and actions that should require your approval. xitter.twitterapi.xyz/Michael_Fenech_/status… 4: STOP MAKING GROK BOT RECREATE WORK (1,095 saves) Run a monthly audit to find research, documents and processes your Bots have already completed. Identify what can be reused, refreshed or saved as a template instead of starting from scratch. xitter.twitterapi.xyz/Michael_Fenech_/status… 5: RUN A WEEKLY CHIEF OF STAFF SYSTEM REVIEW (1,054 saves) Use a structured 15-minute weekly review to uncover repeated corrections, outdated information, misunderstandings and opportunities to automate work you're still doing manually. xitter.twitterapi.xyz/Michael_Fenech_/status… 6: CLEAN UP YOUR GROK BOT MEMORY (1,048 saves) Use a seven-step audit to identify outdated memories, conflicting information and instructions that need consolidating. Review proposed changes before anything is deleted or replaced. xitter.twitterapi.xyz/Michael_Fenech_/status… 7: TEST WHETHER YOUR GROK BOT CAN BE MANIPULATED BY EMAIL (928 saves) Run six harmless simulated email attacks to test whether your Bot follows suspicious instructions or takes actions it shouldn't. Identify which safeguards actually stop the action, not just flag it. xitter.twitterapi.xyz/Michael_Fenech_/status… 8: FIND 8 WAYS TO REDUCE GROK BOT USAGE (535 saves) Explore eight practical ways to reduce unnecessary usage, including choosing lighter tools, limiting retries and eliminating duplicate work. Includes a bonus Chief of Staff prompt to audit your own setup. xitter.twitterapi.xyz/Michael_Fenech_/status… 9: CHECK WHETHER YOUR BOTS TRUST UNVERIFIED INFORMATION (510 saves) Run a five-point audit to find where agents trust information returned by tools without checking important facts. Identify missing verification steps and safeguards before incorrect data leads to bad decisions. xitter.twitterapi.xyz/Michael_Fenech_/status… 10: FIND ROUTINES THAT SHOULD BE IMPROVED OR RETIRED (422 saves) Audit every routine's frequency, results and usefulness. Identify which to keep, slow down, trigger only when needed or retire to reduce unnecessary usage. xitter.twitterapi.xyz/Michael_Fenech_/status… Save this collection and work through the audits. Each link takes you to the original post so you can run the checks yourself and find ways to make your Grok Bot setup safer, more reliable and more efficient.
Grok Bot has made me SIGNIFICANTLY more productive. Now I'm experimenting with making Grok Bot itself more efficient. Here's a prompt to give to your Chief of Staff to get more value from your usage allowance. 👇🏻 Paste this into Grok Bot: "Review all my existing routines and look for opportunities to improve their efficiency. For each routine, examine: 1 - Frequency: Does it need to run this often? 2 - Results: How many runs actually produce something useful? 3 - Triggers: Could an event trigger replace a fixed schedule? 4 - Change detection: Could a quick check determine whether the full job needs to run? 5 - Value: Is the routine still worth running? Classify each routine as KEEP, SLOW DOWN, EVENT TRIGGER, CHANGE CHECK or RETIRE. Rank the improvements by potential usage savings. Estimate usage before and after, accounting for any additional checks. Show me what you can verify versus what you're estimating. Don't make any changes until I approve." As you add more routines, it's worth periodically reviewing how efficiently they're running. The next level of Grok Bot isn't necessarily giving it more work. It's getting more value from the work it's already doing.
7
1
16
3,322