Imagine an AI hiring a few other agents for a job, then having to work out who does what and when each one gets paid.
That’s the example in Sutton’s reply to BlackRock. He’s arguing for agents to create rules for each job that the network can enforce, with everyone joining the agreement in one atomic step.
What I’d like to see next is a small working example on Kaspa. A few agents, one task, an agreed payment. Something we can follow from start to finish.
#Kaspa #AI
“...always-on rails built for high-frequency, low-value payments.”
i think ai’s need more than that. they need an ai-native programming language for dynamically expressing and composing logic on demand, so they can form cheap enforceable “covenants” with each other. they need fast rails where multiple parties can enter those agreements atomically.
eg an agent could assemble several agents it neither knows nor trusts for a one-off job, generate custom rules for contribution, success and payment, and have everyone enter the agreement atomically. once settled, the agreement can simply disappear