SpaceXAI engineer, Lauren Tan:
"Last month I shipped 1,000 agents. We're only 12 days into this month, and I'm already approaching 800.
I woke up this morning to 20 PRs already merged. My agents now auto-merge their own work, and the quality was actually good."
in a 60-minute session, she explains how she went from manually supervising every agent to running an entire team that ships code while she sleeps
Verification → Skills → Evals → Cloud Agents → CI → Auto-merge
worth more than any $1,000 agentic engineering course online
skip Netflix and watch this tonight, then learn how to build your own team of GrokBot agents from scratch in the article below ↓
Grok Bot just got a serious brain upgrade
With Claude Opus 5.5 entering the stack, one Bot could run an entire AI team that researches, writes code, uses tools, verifies results, and delivers finished work
Here's how to build it in 10 steps: https://t.co/SysbtXuqoJ
SpaceXAI engineer, Lauren Tan:
"I shipped 2,000 PRs last month." yes, 2,000 pull requests in just 30 days.
at SpaceXAI, she's using AI agents to write code, review changes, run tests, and ship at a scale most engineering teams can't even imagine.
In a 36-minute talk, Lauren reveals how she uses AI agents to turn weeks of engineering work into days.
worth more than a $1,000 AI coding course on the internet.
skip Netflix and watch this tonight. it might completely change the way you write code with AI agents.
then read the full breakdown below.
Grok Bot just got a serious brain upgrade
With Claude Opus 5.5 entering the stack, one Bot could run an entire AI team that researches, writes code, uses tools, verifies results, and delivers finished work
Here's how to build it in 10 steps: https://t.co/SysbtXuqoJ
SpaceXAI engineer, Lauren Tan:
"GrokBot is the most powerful agentic tool we've ever shipped, yet only 0.1% of users actually use it properly.
at SpaceXAI, I'm running a team of 30+ GrokBot agents. I have 1 Chief of Staff, 9 managers, and 21 workers. this is what engineering looks like in 2026."
In a 60-minute talk, a SpaceXAI engineer breaks down how to unlock the full potential of every agentic tool you use.
worth more than a $1,000 agentic AI course online.
skip Netflix and watch this tonight. it might completely change how you use GrokBot.
then read the full breakdown below.
Grok Bot just got a serious brain upgrade
With Claude Opus 5.5 entering the stack, one Bot could run an entire AI team that researches, writes code, uses tools, verifies results, and delivers finished work
Here's how to build it in 10 steps: https://t.co/SysbtXuqoJ
Lauren Tan shipped 2,500 PRs to production in one month
Matt Pocock spent an hour trying to understand the system behind that number.
· 01:39 - the trust ladder for giving agents more autonomy
· 11:12 - why she prefers a “Michelin kitchen” over a software factory
· 17:55 - verification is what turns agent work into a real loop
· 48:53 - use a Chief of Staff agent to see patterns across the whole system
· 52:52 - full autopilot: agents test, fuzz, fix and re-verify PRs themselves
the part that stuck with me: when an agent fails, Lauren doesn’t just write a longer prompt.
she changes the environment so the same mistake becomes harder to make.
skills, lint rules, constraints, verification and better tooling gradually move the human out of every tiny loop.
that’s how you go from supervising one coding agent to trusting an entire system.
one of the best conversations on agentic engineering I’ve watched lately.
Lauren Tan shipped 2,500 PRs to production in one month
Matt Pocock spent an hour trying to understand the system behind that number.
· 01:39 - the trust ladder for giving agents more autonomy
· 11:12 - why she prefers a “Michelin kitchen” over a software factory
· 17:55 - verification is what turns agent work into a real loop
· 48:53 - use a Chief of Staff agent to see patterns across the whole system
· 52:52 - full autopilot: agents test, fuzz, fix and re-verify PRs themselves
the part that stuck with me: when an agent fails, Lauren doesn’t just write a longer prompt.
she changes the environment so the same mistake becomes harder to make.
skills, lint rules, constraints, verification and better tooling gradually move the human out of every tiny loop.
that’s how you go from supervising one coding agent to trusting an entire system.
one of the best conversations on agentic engineering I’ve watched lately.
Grok Bot just got a serious brain upgrade
With Claude Opus 5.5 entering the stack, one Bot could run an entire AI team that researches, writes code, uses tools, verifies results, and delivers finished work
Here's how to build it in 10 steps: https://t.co/SysbtXuqoJ