Anthropic Engineer Andrej Karpathy:
"The biggest mistake in AI right now - people are forcing agents to work instead of mastering the model first
We made that mistake in 2016 at OpenAI - It cost us 5 years "
what Karpathy actually means:
step 1 → stop forcing your agent to do everything, understand the model underneath first
step 2 → demos are easy - products take a decade. self-driving proved it - if you skip the foundation, everything breaks
step 3 → the agent is not the product. the foundation is. build that - and agents emerge on their own
"you building agents right now - you're at the forefront. not OpenAI. not DeepMind. you "
watch - bookmark, then read article below ↓
Jev Engineering is what happens when agents stop using LLM calls for every tiny decision.
It doesn't write text. It returns typed decisions and calibrated probabilities your code can act on directly
~100 ms per call, $42 per billion tokens, and Vercel's CEO and CTO are already calling it extraordinary
in a 10-minute breakdown, Rob Shocks shows how it works live in the playground and gives the real speed and cost numbers, with the caveats
key moments:
2:08 - real speed, cost, and why this opens up so many use cases
3:31 - the three question types: Choice, Score, Noul
4:23 - where deterministic software meets AI agents
watch it no matter what, then start building with it
Jev could become the control layer AI agents have been missing.
Instead of spending 5–20 seconds and expensive LLM calls deciding every next step, it can route actions in milliseconds at near-zero cost.
In this article, I break down how https://t.co/T9wU8CdOsj
Anthropic engineer:
"99% of people use Claude Code like Google, and only 1% are running swarms of self-learning Claude agents
I'm running 100+ agents in a loop. I have Chief agent, PM agents - they manage the whole team"
In this 30-minute workshop, an Anthropic engineer shows a setup with 100+ agents running continuously
this is worth more than another $500 vibe-coding course
Watch the workshop, then read the article below on how to build self-improving agent systems with Fable ↓
Jev could become the control layer AI agents have been missing.
Instead of spending 5–20 seconds and expensive LLM calls deciding every next step, it can route actions in milliseconds at near-zero cost.
In this article, I break down how https://t.co/T9wU8CdOsj
Annie Wang (Senior Google Developer) spends 15 minutes showing how to wire up the Agent Development Kit
What is ADK → The agentic stack → How agents call tools → Live demo → Deploy
0:39 → Why you need an agent at all
4:20 → The full agentic stack
5:01 → A minimal agent in code
12:56 → Live run in the ADK web UI
One of the few guides that shows real code and architecture - from an empty project to a working MCP agent