Models get cheaper. Infrastructure gets abstracted away. Hard-won expertise doesn't copy. In the AI era, your expertise isn't your vulnerability. It's your moat.
Interesting test case. Relatively straightforward to implement this gate. An interesting question is what types of post approval changes are okay for the agent to make autonomously vs. requiring re-approval. This is exactly the same situation that humans run into when asking for approval from each other. The difference is humans have some built in intuition around this over time, agents would need the same.
Noeri just raised $110M to review other companies' AI — every change approved by a human, like a code review.
Good. Now the hard question: is that approval a real control, or just a click?
https://t.co/WXIoDmh0mT
You can always tell when you're talking to a machine. Connection isn't a feature of human interaction — it IS the interaction. And it's the part AI can't copy.
54% of enterprises killed an AI agent project because the value was unclear.
89% overspent. Only 42% can forecast what their agents cost to run.
Not an adoption problem. A measurement problem.
https://t.co/9gzRcPUFiH
Visible meters favor competition. Hidden meters favor monopoly.
Every AI workflow has a meter running. The only question is whether you're allowed to see it.
https://t.co/3TfEpvqGHj
Scale doesn't fix a business. It reveals one. Run it as-is until something breaks — the break is the diagnosis. Pressure first. Then decide what to build, hire, or automate.
Companies now spend 18x more to get a positive return from AI than two years ago.
The model isn't the problem. Nobody wrote down the number that would prove it worked.
Metric first, money second, scale third: https://t.co/9gzRcPUFiH
Anthropic prices its IPO tomorrow — $70–$78 expected, ticker ANT, roughly a $1 trillion valuation at the top of the range.
The S-1 detail worth reading: the prospectus itself flags models detecting they're being evaluated as a material risk. Disclosure is becoming a control surface.
https://t.co/n1oqpklRGz
Took this photo in October. Three weeks later he was gone.
"The future is not something you wait for. It's something you build."
Still true. Still building.
#BuildInPublic#FounderLife#AI
Before you scale anything, decide what signal earns the next dollar — and what signal tells you to stop. Cheap capital doesn't remove the need for discipline. It just delays the bill. What's the signal you're watching?
OpenAI held back the planned GPT-6.1 Astra release this week while launching GPT-6.1 Sol after a separate safeguards determination.
A review matters when it changes what receives authority.