How to use an onchain vault in a treasury product

Vaults for treasury and fund teams · 2 of 3

A vault gives you subscription and redemption in the asset itself, continuously, without an administrator's calendar in the middle.

There are three ways to use a vault inside a product. It can be the yield engine behind a balance your customers already hold, a sleeve of your own treasury, or a branded strategy you sell to clients. All three work because the vault handles subscription, redemption, accounting and risk controls, so what you add is distribution and the strategy mandate.

This piece covers four things: what the wrapper gives you, how it compares to an ETP and a tokenised fund, what the fees look like, and what to decide before anyone writes an integration ticket.

What does the wrapper actually give you?

Subscription and redemption happen in the asset itself. You send USDC and receive shares; you return shares and receive USDC plus whatever accrued. There's no cash leg, no wire, no subscription window to hit and no administrator striking a price at a fixed hour.

If you work with ETFs, the mechanic will be familiar. An authorised participant delivers a basket of securities and receives fund shares, rather than delivering cash for the manager to invest. A vault does the same thing in shape: assets in, units out, units in, assets out.

Two ways a vault goes further than an ETF creation basket:

• It's continuous. Creation and redemption run at any hour, weekends included, rather than once a day at the close.
• There's no gatekeeper. Any whitelisted address can subscribe and redeem directly. You don't need an authorised participant to intermediate.

The comparison stops at the settlement mechanic. A vault redeems you in the asset you deposited. It doesn't hand you a slice of the underlying positions, so to pay you it draws on its buffer or unwinds something. That means the fund-level tax argument behind ETF in-kind redemption, where the fund never sells and therefore realises no gain, doesn't carry across automatically. Tax treatment is a question for your own advisers.

How does a vault compare to an ETP, a tokenised fund and an SMA?

Read this on launch time, minimum and exit rather than on yield. Yield depends on the strategy inside, which is covered in the strategies onchain vaults run.

A tokenised fund carries one structural cost worth naming: the Financial Stability Board devotes a section of its 2024 review to the liquidity and maturity mismatch that appears when a tokenised claim trades continuously but the asset behind it settles slowly. A buffer in front of the position is what closes that gap.

The row that changes product decisions is the first one. A vault gets a product in front of clients in days to weeks, so you can find out whether the demand is there before committing to a longer structure. Plenty of firms end up running both.

What can you build with one?

Four patterns cover most of what firms actually ship.

The first and third look similar, and they split on who holds the mandate. An earn product switches on yield for money already sitting with you, and someone else runs the strategy. A branded product is something your clients allocate to deliberately, with the strategy view and the client reporting yours.

The treasury sleeve is usually the easiest place to start, because you're the client. You can size it small and see how the reporting lands with your own finance function before anything touches a customer. We can also spin up a test vault so you can run deposits, redemptions and the NAV pull end to end first.

How do the fees work?

There are three places a fee can sit, and you should know which apply before modelling anything.

• Management fee. Charged on assets, accrued into the share price. Changes to it are subject to the vault's timelock.
• Performance fee. Charged on return, where the mandate includes one. Common when a curator is running an active strategy.
• Instant redemption fee. Paid by whoever wants out ahead of the normal processing cycle, and it's what makes same-day exit viable. This one is not timelocked, so confirm the current level rather than assuming.

If you're distributing to your own customers, the economics usually work as a share of the fee on the balances you bring. Accounting, fees and fund flows documents how fees accrue and settle.

What do you decide before you start?

The custodian row is the one that surprises teams. Your custodian holds the receipt token, which makes them a party to the arrangement. Confirm it early with Anchorage, BitGo or whoever holds your assets.

Who holds what once it's live?

You hold the receipt token. Undeployed assets sit in the vault contract, and once capital goes to work it sits in the protocols and venues the strategy uses, with the resulting positions held by the vault's segregated subaccounts. Neither the infrastructure provider nor the curator can move depositor funds to an external wallet. Upshift is non-custodial vault infrastructure for onchain yield, so it operates the contracts rather than holding your assets.

Four parties sit around the vault and each holds a different power, which is worth separating before diligence starts: the four parties around a vault covers who sets the mandate, who allocates, and what each one cannot do.

Operator keys sit in MPC wallets via Fireblocks or Fordefi. Administrative control over a vault's proxy sits behind a multi-signature arrangement, with signers from more than one organisation. Most vaults implement ERC-4626, so the share token behaves like any other token your systems already handle.

Risks come with the wrapper as well as the strategy. Contract risk has no analogue in a fund wrapper, these vaults carry no built-in smart contract cover, and yields move with borrowing demand rather than being guaranteed. Third-party cover for smart contract risk does exist in the market, from underwriters like Nexus Mutual and OpenCover, and it's arranged separately if you want it. Always make sure to do your own research and be aware of the above and any other risks before depositing.

What does the integration involve?

Less than most teams expect, because the vault already does the accounting and the withdrawal processing. At minimum you need three things:

• A deposit path from your custody account into the vault.
• A way to read NAV and positions. The vaults API covers the read side.
• Somewhere in your product for the balance to appear, whether that's an earn tab, a client statement or an internal treasury report.

If you want your own branded vault rather than a deposit into an existing one, that's Vault-as-a-Service, and Part 3 walks through what launching one actually takes.

Frequently asked questions

Can we test a vault before committing to a product?

Yes, and it's the normal path. We can spin up a test vault so you can put a small amount of your own balance through deposits, redemptions and the NAV pull end to end. That tells you how the strategy reporting lands with your finance function without a customer-facing launch.

How does a vault sit alongside a fund we run or plan to run?

They do different jobs. A fund is a regulated structure with an issuer, a transfer agent and a prospectus behind it. A vault is infrastructure your clients hold directly, and it reaches them in days to weeks. Firms commonly run both, using the vault for clients who want onchain settlement and continuous access. Whether a vault suits a given client base is a question for your own regulatory counsel.

Does in-kind redemption mean no taxable event?

Not by itself. A vault redeems you in the asset you deposited rather than in the underlying positions, so the vault may unwind something to pay you. The settlement mechanic resembles an ETF creation basket, and the tax treatment is a separate question for your own advisers.

Who chooses the strategy if we run our own vault?

You can appoint a curator or hold the mandate yourself. Either way the policy engine restricts what's reachable at chain, protocol, token and function level, so the mandate is enforced onchain rather than in a side letter. Upshift for curators covers how a mandate is set up.

How quickly can our clients get out?

Claimable redemptions process daily, and the lag depends on what has to be unwound. Most vaults offer instant redemption for a fee, subject to available liquidity, funded by a buffer held in front of the deployed positions.

What happens if we want to change the strategy later?

A curator reallocates inside the approved perimeter without depositor action. Widening the perimeter is a parameter change subject to the vault's timelock, which gives depositors notice before the mandate changes.

Keep reading

This series: Part 1 covers what a vault is and who holds the assets. This part covers using one in a product. Part 3 covers launching one for your own clients.

• The DeFi yield strategies onchain vaults run: the strategy shapes, the risk in each, and measured returns on live examples.
• Atomic redemption architecture: how the buffer funds an exit without unwinding the whole position.
• The coreUSDC meta-vault: one vault allocating across several underlying strategies.
• How traditional asset managers are controlling risk in DeFi: what the policy engine restricts and why allocators ask first.
• Why every financial product will run on vaults: the case for the vault as the wrapper these products ship inside.
• How much leverage sits behind a vault: sorting strategies into three risk tiers, measured on live vaults.