This article is based on Sentora's Report: The Smart Vault Paradigm. Download it here.

The lending protocols that now sit at the core of DeFi can act in a fraction of a second. The people who design strategies on top of them cannot. 

Most decisions about exposure, parameters, and emergency responses still flow through committee discussions, group chats, and wallets that require several signatures before anything reaches the chain. The gap between how fast these markets move and how fast the humans overseeing them can react has become a serious source of risk on its own.

This article looks at what can be called the Multisig Bottleneck. The name refers to the multi-signature wallets that risk teams use to approve actions, but the broader idea is the speed limit imposed by human coordination on a system designed to run at the speed of code. 

Two recent incidents make the cost of that speed limit clear. The first is the late-2025 liquidity crisis involving the stablecoin xUSD and the platform Stream Finance. The second is the March 2026 exploit of the Resolv protocol. Both point to the same underlying pattern.

The Lending Layer Has Outgrown the People Overseeing It

Older DeFi lending protocols pooled everyone's funds together and let risk teams adjust a handful of settings every week or so. The action surface was small, so a slow review process was good enough.

Newer protocols, including Morpho Blue, Euler v2, and Kamino v2, work differently. They split lending into many separate markets. Each market has its own collateral type, its own price feed, its own borrowing limits, and its own pool of liquidity. A single curated vault, which is the product allocators usually deposit into, might spread money across fifty of these markets or more.

As a result, the number of decisions to track has grown enormously, while the time available to make each decision has shrunk to the length of a single block. Human oversight has not scaled with that change. The contracts on-chain run in seconds, but the people supervising them still rely on group chats, conference calls, and signatures collected one by one.

The Gap Between What the Market Is Doing and What the Team Can Do About It

Sentora describes this as the compute gap. It is the distance between what the market is doing right now and what the team responsible for managing exposure can actually do about it within a useful time frame.

To act, a risk team has to see the event, get on a call, agree on a response, write the transaction, gather signatures from each multisig signer, and broadcast on-chain. Every one of those steps takes time. 

In a quiet scenario, the delays are invisible. But in stressed conditions, the same delays become the reason the response arrives too late to be useful. Safety procedures designed to protect users from rushed decisions end up creating windows during which the protocol is exposed and no one can intervene quickly enough.

This is the heart of the Multisig Bottleneck: Protocols run at the speed of code, but oversight runs at the speed of humans. The mismatch sits in the architecture itself, and every well-run team encounters it.

Withdrawal Traps: When the Team Cannot Move Fast Enough

The late-2025 liquidity crisis showed what this looks like in practice. The stablecoin xUSD lost its peg, and the platform Stream Finance came under heavy pressure. xUSD was being used as collateral across many curated vaults, so the shock spread quickly through the lending ecosystem.

In one sense, the modular design worked. The damage was contained to specific isolated markets and did not engulf the protocols as a whole. The human response, however, could not keep up. As xUSD's market price diverged from the prices the protocols were using internally, automated traders moved to withdraw the available liquidity from curated vaults before the risk teams behind those vaults could meet and react. Users who tried to withdraw their deposits found that the cash had already left.

The teams running these vaults were operating on a slower clock than the actors moving against them. The cause was timing, and the architecture made faster reaction structurally difficult. The protocols themselves continued to function as designed. The oversight layer simply could not match the speed at which the markets were transitioning between states.

The Resolv Anomaly: When the Window Is Open at the Wrong Moment

The Resolv exploit in March 2026 illustrates the same problem from a different angle. An attacker found a way to mint eighty million units of the stablecoin USR without backing them.

The transaction was striking: a deposit of around one hundred thousand dollars produced fifty million dollars in newly minted tokens. To any automated monitoring system, that ratio is an obvious red flag. It should have triggered an immediate response.

To human monitoring, it was effectively invisible. Risk teams check dashboards at intervals and watch alerts, but they do not have the capacity to observe every block continuously. The Resolv exploit took place in the gap between two of those checks. 

By the time the secondary market price of USR began to fall and pulled human attention back to the situation, the damage was already spreading into other parts of the lending system. Fluid, another protocol that had been lending against USR-related positions, absorbed bad debt as a result.

The settings at Resolv and Fluid were within the range the system was designed to support. The cause of the loss was timing: no one was watching at the moment that mattered. In a fast-moving environment, security depends on the cadence at which the system checks itself. Parameters defined on paper are only one part of the picture. The live response cadence is the other.

From Checking Dashboards to Continuous Verification

The xUSD and Resolv events are not isolated incidents. They are the first visible examples of a pattern that becomes more likely the faster protocols get and the more markets they support. Every extra second between observation and action becomes a window someone else can use.

The structured answer is to change how oversight works. Instead of a team checking dashboards at intervals and meeting when something looks wrong, the vault itself should evaluate its state on every block, against a clear set of rules the team has agreed in advance. When a rule is breached, the vault acts. The human role moves up the stack: from approving each transaction by hand to setting the policy that governs what the vault is allowed to do.

This is the architecture Sentora believes the next phase of DeFi requires. The protocols already run at the speed of code. The risk and oversight layer has to operate at the same speed. Capital allocated into these protocols should sit behind a system that watches every block, evaluates exposure continuously, and acts within the same block when something crosses a threshold.

A Disciplined Path Forward

The Multisig Bottleneck is what happens when a high-resolution, second-by-second financial system is overseen by committees. The xUSD episode showed the failure mode under speed pressure. The Resolv episode showed the failure mode under attention pressure. Both will keep happening as the lending environment grows.

Closing the gap means building an oversight layer that operates on the same clock as the protocols it covers. It means moving from manual approvals to structured policies that can be checked and enforced automatically. It means monitoring that does not blink. This is the direction Sentora is building toward: institutional-grade allocation infrastructure that pairs the flexibility of modern DeFi with the discipline of a continuous, governed risk process.

The era of weekly risk meetings overseeing block-by-block systems is closing. The platforms that absorb serious capital in the next cycle will be the ones whose oversight layer has moved to where the risk actually lives: into the policy, into the system, and onto the same clock as the markets it serves.