Coadjute built one of the more interesting proptech theses of the last decade: a shared network connecting all parties in a property transaction — buyer, seller, conveyancer, broker, lender, surveyor, HMRC, Land Registry — on a single ledger. The technology was R3 Corda. The year was 2018. The ambition was to eliminate the weeks of delay caused by parties passing documents through email and waiting for confirmations that arrive in different formats from different systems.
The thesis was right. The substrate was wrong. Not because Corda is bad technology — it is well-engineered for enterprise permissioned networks. But because permissioned blockchains solve the coordination problem by requiring every party to join the same network, run the same software, and agree to the same governance. In a multi-party transaction with regulators, the governance negotiation alone takes longer than the technology build.
What A2A changes.
The A2A protocol — Google’s Agent-to-Agent standard, now at the Linux Foundation — changes the substrate. Instead of requiring every party to join a shared ledger, each party runs their own agent. The agents communicate through a standard protocol. Each agent maintains its own state. Coordination happens through message exchange, not shared infrastructure.
A property transaction in the A2A model looks like this: the buyer’s agent publishes requirements. The seller’s agent publishes a capability manifest for the property. The conveyancer’s agent verifies title. The lender’s agent evaluates risk. The surveyor’s agent publishes inspection attestations. HMRC’s agent checks tax status. Land Registry’s agent records the transfer. Each agent speaks A2A. None of them need to join the same network.
The multi-party coordination that Coadjute tried to solve with a shared ledger happens instead through structured message exchange between independent agents. The governance problem largely disappears because nobody needs to agree on shared infrastructure. Each party controls their own agent, their own data, their own keys.
The 2026 version.
The Coadjute problem restated for 2026: what does an agent-native property network look like? A2A-native, EU-continental (not just UK), notary-integrated (because continental property law requires notarial involvement), and compatible with EUDI wallets for identity verification.
Nobody has built this. The proptech sector is still mostly thinking in terms of platforms — centralized marketplaces that aggregate listings and take fees. The protocol-native version — where the network is the message exchange, not a platform anyone controls — is architecturally possible but commercially unproven.
This is a useful case study for a broader pattern: private permissioned blockchains were the 2018 answer to multi-party coordination. A2A agents are the 2026 answer. The coordination happens at the protocol layer, not the infrastructure layer. The implications go well beyond property transactions, but property is where the pain is most visible and the parties most numerous.
Whether A2A actually replaces Corda-style networks for multi-party workflows is still an open question. The technical case is strong. The adoption case depends on institutional willingness to run agents, which is a cultural shift as much as a technical one. Still watching this one.