devFC

I see a bunch of blockchain agents with control over funds. There the control safe guards so far have been in the functions the model call. 

Execution though is instant and irreversible. That allows Agents to autonomously trust each other assuming they can access the transaction status.

With stripe they could do the same according to the APIs I assume; but still have to trust stripe in the end.

This is so smart and practical.

If you are a business where developers can spend / collect money with you:

Add APIs so (AI) bots can do the same… and enable new use cases while increasing revenue!

(And leave it to devs/customers to ensure safeguarding these use cases!)

The big challenge with LLMs / AI agents having access to move money is exactly this: how do you ensure they don’t make mistakes (and lose money) given their non-deterministic nature?

Stripe offers rails: customers need to safeguard their usage (and stand in for losses if failing…

Gergely Orosz

I see a bunch of blockchain agents with control over funds. There the control safe guards so far have been in the functions the model call. 

Execution though is instant and irreversible. That allows Agents to autonomously trust each other assuming they can access the transaction status.

With stripe they could do the same according to the APIs I assume; but still have to trust stripe in the end.

https://x.com/gergelyorosz/status/1857541550684643550?s=46

Today(), @stripe is launching a SDK built for AI agents:

- LLMs can call payment, billing, issuing, etc APIs
- Integrates with @vercel, @LangChainAI, @crewAIInc
- Use any model via functions (w/ per token billing)

Excited what you, and your bots, build: https://t.co/3GiXmNm6en