#newt $NEWT @NewtonProtocol
Been seeing Newton (NEWT) pop up a lot lately, and what they're building is kind of the opposite of flashy. While everyone's hyped about AI agents trading and managing money on their own, Newton is asking a quieter question: who actually checks if that agent should be allowed to do this in the first place?
#Newt
Their setup is pretty simple once you get it. You write rules — using stuff like sanctions lists, identity checks, risk limits — and those rules get checked right before a transaction goes through, not after the money's already moved. newton That's the part I like. Most compliance tools in crypto only kick in after something's gone wrong. This one's a gate, not a cleanup crew.
The bit that really stuck with me though is what they're doing for AI agents specifically. They've added guardrails like spending limits, approved recipients, and protection against prompt injection — all checked before the transaction settles. newton
Think about it this way...if an agent gets tricked by a sketchy prompt into trying to send funds somewhere it shouldn't, that check isn't relying on the agent "knowing better." It's sitting outside the agent, at the contract level, where a manipulated prompt can't talk its way around it. And since every decision leaves behind a signed receipt anyone can verify on-chain, newton you're not just hoping it worked, you can actually check.
Feels like one of those essential pieces of infrastructure that has to exist before "AI agents managing real money" stops being a scary idea and starts being a normal one.
Been seeing Newton (NEWT) pop up a lot lately, and what they're building is kind of the opposite of flashy. While everyone's hyped about AI agents trading and managing money on their own, Newton is asking a quieter question: who actually checks if that agent should be allowed to do this in the first place?
#Newt
Their setup is pretty simple once you get it. You write rules — using stuff like sanctions lists, identity checks, risk limits — and those rules get checked right before a transaction goes through, not after the money's already moved. newton That's the part I like. Most compliance tools in crypto only kick in after something's gone wrong. This one's a gate, not a cleanup crew.
The bit that really stuck with me though is what they're doing for AI agents specifically. They've added guardrails like spending limits, approved recipients, and protection against prompt injection — all checked before the transaction settles. newton
Think about it this way...if an agent gets tricked by a sketchy prompt into trying to send funds somewhere it shouldn't, that check isn't relying on the agent "knowing better." It's sitting outside the agent, at the contract level, where a manipulated prompt can't talk its way around it. And since every decision leaves behind a signed receipt anyone can verify on-chain, newton you're not just hoping it worked, you can actually check.
Feels like one of those essential pieces of infrastructure that has to exist before "AI agents managing real money" stops being a scary idea and starts being a normal one.