He Deployed 20 Round-the-Clock Grok Bots, and One Sold His Old Tesla in 24 Hours
Christian Reber has publicly shared his Grok Bot automation system: 20 agents run around the clock in the cloud, the House Bot sold his old car in one day, and the Chief of Staff outputs a PDF guide. He also answered questions about cost, architecture, and privacy.
Christian Reber publicly shared his Grok Bot automation system on X. Twenty AI bots run around the clock in the cloud. Most work starts with voice notes, recorded in the Grok app on a Mac, iPhone, or Tesla.
Bots take over the tedious tasks, leaving him time for founders. He also had the Chief of Staff Bot prepare a PDF cheat sheet for founders of portfolio companies: interfacecap.com/grok-bot-setup.pdf.
Three accompanying images:
There are more technical details in the replies. On architecture, he explained: each Bot maps to one agent, with no duplicates. When tasks pile up, the Bot spins up background helpers (sub-agents) on its own while keeping itself available.
On cost, he uses Grok Ultra at $200 per month. He has never hit the usage cap or paid extra; the trick is to have the Chief of Staff 'save tokens whenever possible.' One proof point: the House Bot sold his old Tesla for $29,000 within 24 hours, effectively earning back the monthly fee.
Someone asked how to set it up from scratch. His advice: first create a Chief of Staff, upload the PDF, then ask it by voice—'How can this system be used in my business? Which Bots are most critical for automating my work and life tasks?'
On privacy concerns, he was very candid: 'Google is already reading my email, 1Password has all my passwords, Apple has my photos—I've been fully exposed for a long time.' He does not use Dot because it is 'too slow.'
One user said they also use Grok Bot to automate Instagram/TikTok content creation and publishing, with good results; another Ultra user said they would probably use up 100% of their quota by around day 4-5. With the same subscription, usage varies significantly.
The most interesting thing about this system is how many real-world tasks a sufficiently flexible agent network can handle for an individual under an existing model subscription—from everyday information organization to completing a transaction. The details in the PDF may be more specific, but this case alone is worth studying.
发布时间: 2026-11-20 08:18