Holy f*ck. SpaceXAI engineer Jediah Katz (ex-Cursor):
“Within six months, ninety percent of agent work is going to run off the laptop.
The harness provides the runtime, persistent state, and access controls that make long-running agents reliable.”
In this 18-minute talk, he explains why smarter models don't replace the system around them: agents still need persistent execution, permissions, memory, and coordination.
Model → Harness → Tools + Memory → Agent Teams → Work That Runs Without You
A smarter model doesn't keep your laptop awake. A harness gives agents the infrastructure to keep working after you log off.
More useful than most $1,500 agent engineering courses.
Bookmark it and watch today.
Then read the full “Harness, Loops, Graphs” guide below to build your own agent harness ↓
Your agent loop is probably a cron job with your credit card
Researchers checked 36,487 repos. Nearly all loops had a trigger. Far fewer had a check, state or an exit
What's missing across the harness, the loop and the graph ↓ https://t.co/oUriLMbQ69
Holy f*ck. SpaceXAI engineer Jediah Katz (ex-Cursor):
“Within six months, ninety percent of agent work is going to run off the laptop.
The harness provides the runtime, persistent state, and access controls that make long-running agents reliable.”
In this 18-minute talk, he explains why smarter models don't replace the system around them: agents still need persistent execution, permissions, memory, and coordination.
Model → Harness → Tools + Memory → Agent Teams → Work That Runs Without You
A smarter model doesn't keep your laptop awake. A harness gives agents the infrastructure to keep working after you log off.
More useful than most $1,500 agent engineering courses.
Bookmark it and watch today.
Then read the full “Harness, Loops, Graphs” guide below to build your own agent harness ↓
Holy f*ck. SpaceXAI engineer Jediah Katz (ex-Cursor):
“Within six months, ninety percent of agent work is going to run off the laptop.
The harness provides the runtime, persistent state, and access controls that make long-running agents reliable.”
In this 18-minute talk, he explains why smarter models don't replace the system around them: agents still need persistent execution, permissions, memory, and coordination.
Model → Harness → Tools + Memory → Agent Teams → Work That Runs Without You
A smarter model doesn't keep your laptop awake. A harness gives agents the infrastructure to keep working after you log off.
More useful than most $1,500 agent engineering courses.
Bookmark it and watch today.
Then read the full “Harness, Loops, Graphs” guide below to build your own agent harness ↓
Your agent loop is probably a cron job with your credit card
Researchers checked 36,487 repos. Nearly all loops had a trigger. Far fewer had a check, state or an exit
What's missing across the harness, the loop and the graph ↓ https://t.co/oUriLMbQ69
This is f*cking insane. Lydia Hallie from Anthropic's Claude Code team just explained how Anthropic went from 2,500 code changes a day to more than 30,000:
"If you do want the guarantee, you can only enforce that in the harness"
In this 17-minute workshop, she explains how their sessions changed. Engineers no longer approve every command, they run dozens of sessions for hours and act as their managers
Most of last year's tricks now hold the models back. Shouting IMPORTANT in CLAUDE.md just makes them go off track
What still matters is the harness: deny rules, a sandbox, hooks that run the tests, CI that checks every PR
Old Habits → Better Models → Harness → Dozens of Sessions Running Without You
Everyone's still tweaking their prompts, Anthropic moved the work into the harness
Bookmark and watch it today
Then read how to build that harness in the guide below ↓
Harness, loop and graph could be the missing layers between AI models and agents that actually work.
Same model, same prompt, completely different results.
In this article, I show you how to build all three layers. https://t.co/W99aPfcSub