#newt $NEWT @NewtonProtocol
The more I look at Newton’s Keystore, the less I think the real story is “AI trading.”
The interesting part is control.
A trading agent does not need full access to your wallet. It might only need permission to swap one token, on one venue, within a price range, with a hard daily limit, for a few hours. That sounds obvious, but most automation systems still ask users to trust far more than the task actually requires.
Newton’s approach is trying to make those limits explicit and verifiable. That matters because the biggest risk in on-chain automation is not whether the bot works. It is what happens when it behaves badly, the market moves fast, a key leaks, or a policy needs to be revoked immediately.
The Keystore model looks promising, but it also moves complexity into new places: tree construction, permission updates, operator reliability, and revocation speed. So the real test is not whether Newton can automate trades. It is whether it can keep authority tightly scoped when conditions are messy.
That is the angle I find compelling: useful agents should not need powerful wallets. They should need precise permissions.
The more I look at Newton’s Keystore, the less I think the real story is “AI trading.”
The interesting part is control.
A trading agent does not need full access to your wallet. It might only need permission to swap one token, on one venue, within a price range, with a hard daily limit, for a few hours. That sounds obvious, but most automation systems still ask users to trust far more than the task actually requires.
Newton’s approach is trying to make those limits explicit and verifiable. That matters because the biggest risk in on-chain automation is not whether the bot works. It is what happens when it behaves badly, the market moves fast, a key leaks, or a policy needs to be revoked immediately.
The Keystore model looks promising, but it also moves complexity into new places: tree construction, permission updates, operator reliability, and revocation speed. So the real test is not whether Newton can automate trades. It is whether it can keep authority tightly scoped when conditions are messy.
That is the angle I find compelling: useful agents should not need powerful wallets. They should need precise permissions.