Last year, a health check found that your blood sugar was high. The doctor said it had been above the limit three months earlier, but nobody told you back then. The data on the report had been sitting there all along—no one watched it for you, and by the time you realized it yourself, three months had already passed. Traditional DeFi risk management follows the same logic. Oracle prices are only used to calculate liquidation thresholds; by the time liquidation happens, you’ve already lost money. The trade has already been executed, the risk exposure has already been exposed, and then you rely on liquidation mechanisms to make up for it.
@NewtonProtocol Newton flips this timeline. Before every trade can enter the settlement phase, it must first pass through the strategy engine. The Rego rules say no—so the signature won’t be produced, and the transaction can’t execute at all. The rules are written in Rego—the same kind of thing Goldman Sachs and Capital One use to review the movement of tens of trillions in capital. Newton ports it on-chain, working with a decentralized operator network to generate BLS signature authentication, with on-chain contracts handling the verification.
June 23 Mainnet Beta went live, and RedStone’s feed price and Credora’s risk rating were connected into the strategy engine. Over at the vault manager, you set a rule: “The collateral ratio must not fall below 150%.” Each time someone tries to withdraw or borrow, Newton first asks RedStone for the current asset price and compares it against the preset collateral ratio. If the price drops below the red line, the transaction is blocked right at the door—no need to wait for settlement. Every blocked transaction generates an on-chain certificate that records the reason it was stopped. You can check it on the Newton Explorer—anyone can verify it, and no one can alter it.
This “pre-approval, then settlement” design is the same as the authorization logic used by Visa credit cards. When you swipe a card, the Visa network performs an authorization check before the payment is completed; only if it passes does the transaction go through. Newton brings the same logic onto the blockchain—before settlement, every transaction must pass through the strategy engine, and only compliant ones are allowed through.
For scenarios where AI agents automatically execute actions, this kind of “pre-trade risk control” is almost the only practical risk management approach. When AI initiates trades at machine speed, human post-hoc review simply can’t keep up. Blocking before execution is an entire alert cycle faster than assigning blame afterward. The lab test report tells you what went wrong in the past three months—Newton drops the gate the moment the problem occurs.