News for hackers. Stuff that matters.

ArielBarack

I actually like the intern analogy. The gap is that interns spend at human rate and you can intervene; agents can make thousands of spend decisions/hour and retries/loops can multiply. So I’m trying to understand what people use as the “expense policy system + approval workflow + revocation” equivalent when the spender is software. What’s your practical setup for that today?
This is super helpful — thank you. Two questions:

Did you keep “credits” purely internal, or do you reconcile them back to real invoices/costs per provider?

Any edge cases you hit around retries/idempotency (e.g., task retries after deduct, provider call succeeds but worker crashes, etc.)?

How are people handling paid external APIs for autonomous agents?

I’m curious how people are actually doing this in production, not in demos.If you’re running autonomous or semi-autonomous agents that:call paid APIspurchase datainvoke metered toolschain actions without human approval…how are you handling payments and limits?What I mostly see in the wild are workarounds:provider-level API keys with hard usage capsproxy services that re-bill...

( Read More... | 4 comments | on HN )
Tools of titans, Tim Ferris.

Body and mindset hacking from leading thinkers and experimenters, in a concise format

Re:Jawbone denies 'abandoning' customers on Tuesday March 07, 2017 @07:13AM (#13809776)
I can confirm the article is spot on, been chasing their support since November and still no reply
Re:Ask HN: What podcasts are you listening to? on Thursday February 16, 2017 @03:19AM (#13658412)
the tim ferris show - incredible insights by interesting people