Operations

How Miles Deutscher Runs AI Edge with a Team of 9 Grok Bot Agents

Built by Miles Deutscher · · Source: X

Quick answer

Miles Deutscher built 9 Grok Bot agents for their media and tech business, AI Edge, split into General, Marketing, Product and Sales. A Chief of Staff bot hands out the work and sends a morning brief and an evening summary.

Result: According to Miles Deutscher, the team saves about 53 hours a week (roughly $138,000 a year at $50/hr) for about $4,800 a year in costs, which they put at around a 29x return.

What they built

Miles Deutscher runs AI Edge, a media and tech company with about 50 people. They set up nine Grok Bot agents and split them into four departments, the same way you'd organise human staff:

  • General: a Chief of Staff that leads the team, plus an Executive Assistant for admin like tasks, attendance, email and meeting invites.
  • Marketing: a Creative Researcher that scans X and YouTube for post ideas, an ideation "sparring partner", and a CMO bot that reviews content performance every week.
  • Product: a Head of Product, plus a Funnel Design bot that checks landing pages, sign-up flows, emails and checkout to see where people drop off.
  • Sales: an Outreach bot for partnership reports, inbound messages and outreach drafts, and a Partnerships Management bot for CRMs, deliverables and campaigns.

They say many staff members also have their own bots, and any bot can get subagents for narrower jobs.

How it works

Miles only manages the Chief of Staff. It decides which bot owns a task, passes it on with the right context and follows up until it's done. The bots message each other and share context, so Miles doesn't have to copy notes between chats.

Scheduled routines split the work into a night shift and a day shift. Overnight, Outreach researches 50 possible partners for the BD team, the researchers look for post ideas, and the EA rounds up calendar, Slack and Notion updates. At 7am the Chief of Staff sends one brief that pulls it all together. In the evening it sends a wrap up of what got done, and the CMO, Head of Product and Partnerships bots send their own summaries. They also use native X connections for research and DMs, Slack and Notion, and the iPhone action button to reach bots quickly.

The result

Miles estimates the team saves about 53 hours a week, worth roughly $138,000 a year at a blended $50/hr. They put the cost at about $400 a month, which works out to roughly a 29x return. Their downsides: the first week takes some babysitting, costs can climb, you can't pick the model, and some responses are slow. They give it 9/10 overall.

How to build something similar

  • Start with one coordinator agent, then add specialists only for the parts of the business that eat the most time.
  • Write a short brief for each agent covering what it owns, which tools it can use and what it should hand back.
  • Put recurring work on schedules and have the coordinator roll everything up into one daily summary, so you're reading one message instead of nine.
  • Set a spending limit early and check the output closely for the first week or so.

The original post

https://x.com/i/article/2114838592145268736 I've tested every single AI agent on the market for the past year. Grok Bot is hands down the best. And it’s not close. I now run a team of 9 Bots inside my media and tech business (@aiedge_) and it's genuinely been the highest-ROI thing I've done for the company all year. My bots help my creative team with virality research, my developers with coding, and, as CEO, they help me with everything else - high-level decisions, marketing, product design, etc. If you're a business owner, aspiring entrepreneur, or just someone who wants to make money online, you'll find this article valuable. We'll walk through the entire agentic architecture, a day in the life, the ROI this team has delivered, and my honest take after a full week of testing. Table of Contents I: Why Grok Bot? II: Meeting the Team III: Day in the Life IV: ROI Analysis V: Build Your Own Team (Starter Prompt) VI: My Personal Analysis (pros & cons) I've attached prompts for each of my team members so you can easily spin up your own variations. I: Why Grok Bot? I’ve run almost every agent on the market through one simple test: can it actually do real work inside my business like a human employee without me babysitting it? Most fail, but Grok Bot passed for four specific reasons. 1. It’s built for 𝕏 For my media and tech company, this is the big one. Grok Bot connects natively to all my 𝕏 accounts. I use it for real-time research, news analysis, DM management, etc. (we'll dive into exact use cases later). Bots can search 𝕏 and read posts and profiles through the native 𝕏 connection. No other AI agent has this capability - unless you pay for the API. 2. The UI is so easy Simply put, Grok Bot "just works." The UI is lightweight and easy to navigate. Whenever I need to set something up, Bot can easily create pop-ups, visuals, and more. I message my Bots the same way I message my team. 3. The Chief of Staff setup No other agentic tool has gotten this right. You can have one Bot sit on top as Chief of Staff with specialist Bots underneath. They message each other, share context, and hand off work, so I’m not the middleman copying notes between chats. 4. The team ships constantly For context: Grok Bot launched in August as version 0.17. It’s now on 0.75. The team ships 100+ features every week. When it comes to building my business on a tool, its improvement speed is everything. Every week, my Bots get noticeably better without me doing a thing. I want a single tool I can rely on for months to come, and I feel confident taking that bet on Grok Bot. II: Meeting the Team I've divided my team into 4 departments; obviously, your needs will vary depending on your business. General Marketing Product Sales Department 1: General This is the "command center." I've personally set up two Bots here. Chief of Staff This is the team lead and my primary Bot. I use it to spin up new bots, make connections, dive into my tools (Slack, Notion), & more. When I hand it a task, it decides which Bot should own it, passes it over with the right context, and follows up until it’s done. I don’t manage 9 Bots. I manage one. 2. Executive Assistant Handles all my admin work. Creates tasks, manages team attendance, pulls emails, sends meeting invites to my team, and so on. Department 2: Marketing For a media company, this is the most useful department. So far, I have three bots. Creative Researcher Scans 𝕏, YouTube, and helps me identify daily post ideas, scripting, and related tasks. 2. Ideation/Sparring Partner My "sparring partner." 3. CMO The head of marketing. It sits above the other two marketing Bots and owns the big picture. Every week, it reviews how our content performed and generally helps me with big marketing decisions for the brand. Department 3: Product We have 4+ products. Info, consulting, and some long-term digital products we're building in the background. These Bots help me map the brand's lon
Miles Deutscher (@milesdeutscher) · October 27, 2026 · View on X

Want to build something like this?

Join the Shipping Skool community free. Paid members get live bootcamp calls every week.

Join Shipping Skool — Free

More use cases

← All AI agent use cases