The Decentralised Intermediary

The Decentralised Intermediary Decentralisation is not the absence of intermediaries. It is a governance regime that makes extraction structurally expensive. Until we internalise that distinction, "build crypto-native institutions" will remain a slogan rather than a strategy.

Three essays published [1][2][3] - by Dean Eigenmann, Evgeny Gaevoy, and Meltem Demirors - all arrive at some version of that call. The diagnosis is shared and correct: what arrived was not adoption but absorption. Meltem's economic receipts make the case clearly. But none of the three specifies what makes an institution structurally crypto-native rather than just staffed by crypto people. Tether is crypto-native. Coinbase is crypto-native. Being crypto-native did not prevent Tether from capturing billions in net interest margin or Coinbase from building a regulatory moat around custody and listing fees.The question is not who runs the institution. It is what governs it. I have spent two years working on a specific answer: the decentralised intermediary [4][5][6].

The boundary problem

Not all essential services can be performed on-chain. Stablecoins need someone to hold the dollars. Lending protocols need oracles that report off-chain prices. Tokenised assets need someone to verify the underlying exists. Cross-chain protocols need solvers to execute user intents. At every boundary between on-chain and off-chain, you find an intermediary performing a function that cannot be fully automated.If that intermediary answers to shareholders under corporate governance, the extraction Meltem describes is not a bug. It is the expected output. And if you make the role permissionless and hope competition solves it, the data says otherwise.In empirical analysis of solver networks [7][8], we found that across months of CoW Protocol order flow, even with open entry, three solvers capture over half of all volume. Most participants operate at a loss. The winners are not the ones with better algorithms but the ones with private liquidity access, vertically integrated infrastructure, and capital advantages that smaller operators cannot replicate. Permissionless access prevents gatekeeping. It does not prevent consolidation.

The decentralised intermediary

The alternative is to accept that intermediaries are necessary but change the terms under which they operate. Replace corporate governance with protocol governance. Replace periodic audits with continuous on-chain transparency. Replace regulatory enforcement with programmatic accountability.Concretely, this means an entity whose relationship to the protocol has four architectural properties:

Continuous licensing. The intermediary earns the right to operate through ongoing performance against protocol-defined standards. Not a licence granted once and reviewed annually, but a permission that exists only as long as obligations are met and measured in real time.

On-chain transparency. Every material action is recorded on-chain: collateral movements, risk parameter changes, transactions processed, exceptions granted. Not reporting fulfilled after the fact but an operational requirement embedded in the workflow.

Programmatic accountability. Rewards for performing well, penalties for performing poorly, exclusion for persistent failure. The protocol does not need to sue. It needs to stop calling.

Engineered contestability. As the solver data shows, permissionless access alone does not prevent consolidation. Protocol-level design must actively counterbalance it. This means shared infrastructure as a public good: common liquidity indexes, neutral routing layers, open simulation tools. It means functional unbundling: separating roles so that no single actor dominates through vertical integration. And it means liquidity transparency: removing private access as a competitive moat.

These four properties can be built today with existing smart contract infrastructure. They constitute the first phase of the decentralised intermediary: governance over what happens on-chain.

But decentralised intermediaries operate at the boundary, and many of the assets and actions they handle exist off-chain. A stablecoin custodian holds dollars in a bank account. A tokenised loan is backed by a real-world receivable. Without enforcement that crosses the boundary, on-chain governance alone leaves a gap that sophisticated actors will exploit.

The second phase closes this gap through bidirectional enforcement: oracles that not only report off-chain data to the chain but also transmit on-chain decisions to off-chain registries. An asset encumbered on-chain should be blocked from transfer off-chain. A default recorded off-chain should trigger liquidation on-chain. The enforcement mechanisms themselves are not new: notaries register liens, custodians place holds on pledged securities. What changes is who triggers them and when. Traditional finance relies on soft self-regulation: regulators delegate oversight to institutions and validate after the fact. Decentralised rails make real-time self-regulation possible - the protocol enforces continuously, not because a regulator asked but because the mechanism requires it.

What does not yet exist is the infrastructure to connect these mechanisms to blockchain states, and the legal frameworks that recognise blockchain records as authoritative triggers.

To make this concrete: consider a stablecoin custodian operating under this regime. Today, Tether holds the reserves, publishes quarterly attestations, and answers to no protocol-level constraint on its behaviour.

Under a protocol designed with decentralised intermediary principles, the custodian's reserve movements would be logged on-chain in real time. Its right to custody would depend on continuous compliance with protocol-defined reserve ratios. Deviations would trigger automatic penalties, and persistent failure would result in the protocol routing to a competing custodian. And through bidirectional enforcement, the reserves themselves would be encumbered: the custodian could not move them off-chain without the protocol's consent. The custodian still earns revenue. But the surplus it can extract is bounded by constraints that reach into both worlds.

What this does not solve

Protocol governance itself requires careful design. Who sets the standards? Who updates the parameters? The governance-of-governance problem is real, and I will not pretend it is solved. But the decentralised intermediary framework does not require perfect protocol governance. It requires protocol governance that is better than the status quo of no governance at all over the entities that operate at the boundary.

There is also the question of incentive. Why would an intermediary voluntarily submit to this regime when it could incorporate offshore and do as it pleases? Because protocol access is the product. A solver that operates within a protocol's governance framework gets order flow. A custodian that meets the protocol's standards gets deposits. The constraints are the price of access to a market that would not exist without the protocol. This is not hypothetical: validators already accept slashing conditions in exchange for staking rewards. The decentralised intermediary extends the same logic to off-chain service providers.

Dean ends his essay with a call for conviction. Meltem ends hers with "believe in something, otherwise we stand for nothing." But conviction does not survive contact with institutional incentives. When nothing structural prevents extraction, extraction happens.

Dean asks for systems resistant to capture by architecture. This is what that architecture looks like: not a protocol with no intermediaries, but a protocol that governs its intermediaries so tightly that capture requires subverting the protocol itself.

References:

[1] Dean Eigenmann, A Return to Fundamentals (Feb 2026)
[2] Evgeny Gaevoy, Golden Path (Feb 2026)
[3] Meltem Demirors, From "Adoption" to Annexation (Feb 2026)
[4] Nadiem Sissouno, Intermediaries: Making or Breaking the Stablecoin Market (2024)
[5] Nadiem Sissouno, Intermediaries: Making or Breaking the Stablecoin Market - Part 2 (2024)
[6] Nadiem Sissouno, Intermediaries: Making or Breaking the Stablecoin Market - Part 3 (2024)
[7] Sprinter, Building Economic Trust in Solver-Based Networks: Part 2 (2025)
[8] Sprinter, Building Economic Trust in Solver-Based Networks: Part 3 (2025)