Decentralised finance spent several years being described as the thing that would replace banks. It has not, and the articles saying it would have aged badly.
What is left after the noise is more interesting: a handful of primitives that genuinely work, a larger set that only works if you accept risks most companies cannot, and a clear line between them. Here is roughly where that line sits.
What DeFi actually is
Financial services running as code on a public blockchain, with no institution in the middle. Lending, borrowing, exchange and settlement happen through smart contracts, which execute when their conditions are met and cannot be talked out of it.
That last part is the whole proposition and the whole problem. No intermediary means no gatekeeping, no opening hours and no counterparty deciding you are not worth serving. It also means no one to call when something goes wrong.
The parts that work
Stablecoin settlement. Moving value between parties in minutes rather than days, without correspondent banking, is real and in production. For cross-border payments in particular the difference is not marginal. This is the piece most businesses touch first, and often the only one they need.
Exchange without an account. Decentralised exchanges let you swap assets straight from a wallet. Fees are usually lower than a centralised venue and you keep custody throughout. For treasury operations at modest size this is genuinely useful.
Programmable settlement. Releasing payment when a condition is met — goods received, a milestone signed off, an oracle reporting a value — removes a reconciliation step and the argument that goes with it. Where the condition is objectively checkable, this works well.
The parts that are still research
Lending and borrowing at business scale. The protocols work. The economics require overcollateralisation, so borrowing $100,000 means locking up rather more than $100,000 in volatile collateral. For most companies that is not financing, it is an expensive way to stay long.
Yield. Liquidity provision and yield farming generate returns that look attractive next to a deposit account, and they are compensation for risks — impermanent loss, protocol failure, token depreciation — that a treasury policy usually forbids for good reason.
Anything requiring identity. Regulated finance needs to know who it is dealing with. Permissionless systems are built specifically not to. The bridges between those two worlds exist and are immature.
The risks that decide it
Smart contract failure. Code holding value is a target, and the history of the space is a history of exploits. An audit reduces the risk and does not remove it. Ask when a contract was last audited, by whom, and whether the report is public — and treat an unaudited contract holding your money as a decision, not an oversight.
Regulatory uncertainty. Rules are arriving unevenly across jurisdictions. A structure that is fine today may need unwinding, so anything you build should assume it will be re-papered at least once.
Volatility. Not just of the assets. Gas costs, liquidity depth and yields all move, and a model built on one week's numbers will not survive the next quarter.
Irreversibility. A transaction sent to the wrong address is gone. Not "difficult to recover" — gone. Any process touching this needs controls that assume a human will make a mistake, because eventually one will.
Where this leaves a business
Three honest positions, depending on what you are trying to do.
If you move money across borders and the delays and fees are a real cost, stablecoin settlement is worth a serious look. The technology is ready and the operational questions are manageable.
If you want programmable settlement between parties who already trust each other, a smart contract can remove reconciliation work. Worth prototyping, provided the trigger condition is genuinely objective.
If you are considering DeFi for treasury yield, the risk-adjusted answer for most companies is no. The returns are real and so are the ways they disappear.
The mistake worth avoiding is treating "we should do something with DeFi" as a strategy. It is a toolbox. Bring a specific problem — a settlement delay, a reconciliation cost, a market you cannot reach through banking rails — and the tools are either right for it or they are not.
We build and audit the contract layer this runs on. If you have a specific case in mind, we will tell you honestly whether a chain earns its place in it.
