I used to think the hard part was building the new system. Cleaner rails, instant settlement, programmable assets — get that right and the rest follows. Then I remembered the old system isn't going anywhere, and the chain doesn't get to replace it. It has to plug into it.
A regulated institution runs on decades of infrastructure — core banking, SWIFT messaging, CSD connections, accounting engines, reporting pipelines — with regulators, auditors and processes wrapped around every piece. A tokenized bond doesn't escape that. Its coupon still has to land in a bank account through existing payment rails. Its position has to appear in the firm's book-of-record. A corporate action has to flow into downstream reporting. If the chain can't speak to that world, the promised efficiency dissolves into manual reconciliation at the seam — the exact work you were trying to delete.
So you get two hard choices: build a thick adapter translating on-chain events into ISO 20022 messages, CSD instructions and ledger entries — expensive and brittle — or ask a bank to rip out core systems for a nascent asset class, which won't happen. Most pilots run the chain as a sidecar and reconcile by hand, which doesn't scale.
This is where Dusk's instinct to work through licensed venues and custodians who already hold the legacy plumbing, and to align with existing standards, is the realistic path. It makes the seam thinner — but it's a translation problem, not a solved one.
Who benefits: institutions where a thin, standards-based adapter genuinely removes reconciliation. What makes it fail: underestimating the integration tax — treating the chain as the finish line when it's one node that has to talk to fifty years of plumbing.