We spent 47k running AI agents in production

Ok, so I'm not a programmer, just a knuckle dragger who is vibe coding myself a digital assistant to help me prioritize emails and do scheduling. And because I may use a cloud AI's API rather than local processing I already consider the following to be essential: Token estimation, agent state persistence, cost monitoring and rate limiting, circuit breakers, retry logic, context caching, deadlock detection.

Those are just some of the requirements for AI agent deployment that this article mentions. And hell, I'd want some of them even if I were running the agents on my own GPU...

How does anyone have the brass stones to go to production without at least the precautions that I found necessary within half a day of thinking about it?

Is there really so much funding in the AI space?

I don't know. The article is simply embarrassing. You've let agents running for weeks before pulling the plug? It's incompetence and stupidity. And if you make this to sell your product to the dev crowd, it's also disrespectful to this audience.

I'm curious to see how quickly the tech leaders change their stance on climate change, when they have to use an unbelievable amount of power for these AI agents/services, which directly contributes to the climate change they've been lecturing us on for years.

Bill Gates already changed his stance.

Try Anthropic’s Claude Agent SDK and let Claude orchestrate the subagents, the websearch and a sandbox to do some statistics.

A computer is all you need.