Treasury used to run on the business day. Here’s what replaced it.

Settlement that used to take up to three days can now complete in minutes.

That one change means you could move money 50 times a day instead of once a week. It also means the business day is no longer the thing treasury is organized around. Nobody decided that. Settlement simply stopped observing it.

That shift has a name: always-on treasury. Here’s what it is, what it isn’t and what it asks of your controls.

The old rhythm, and why it worked

For decades, treasury ran on a reliable clock. Positions were struck at end of day, files went out at cut-off and liquidity was swept on a schedule.

That rhythm matched the rails underneath it, and it gave controls a natural place to sit: between the decision and the money moving.

What has changed is not your team’s judgment. It’s the clock the market runs on.

What always-on treasury actually means

Always-on treasury is often described as a speed upgrade. It isn’t. It’s a different operating model, and it comes from three forces arriving together.

Adaptable solutions gave treasury its foundation. Liquidity, forecasting, risk, netting, reconciliation and payments no longer have to arrive as one big migration. You can start with your greatest need and add the rest as you go.

AI collapsed the time it takes to decide. Forecasting becomes continuous instead of weekly. Exceptions get flagged as they happen, with a proposed fix and the policy clause behind it.

Digital rails are collapsing the time it takes to act. When settlement runs around the clock, liquidity can’t sleep. Positions have to be in place before they’re needed, not after the next business day opens.

Separately, three trends. Together, one shift.

Why AI and digital need each other

Here’s the part that rarely gets said out loud. Moving money 50 times a day creates far more opportunity. It also creates far more work.

No treasury team was staffed for that volume, and few want to be. AI makes it manageable. It doesn’t replace the work; it makes the new volume survivable.

Neither one is much use to you without the other. Faster rails without faster decisions just create a faster bottleneck. Faster decisions without a way to act on them stay recommendations.

What always-on treasury looks like on a Tuesday

The clearest way to see the shift is a single day. Here is what a treasury day can look like when it runs on the moment, with each step prepared or carried out at the autonomy you’ve set:

• 03:40 Statements read across every account. Late files flagged.
• 04:12 Overnight sweep of idle and operating cash executed under your investment policy and your bank instructions.
• 05:58 Reconciliation complete.
• 06:30 Entity funding prepared before balances fall short, or carried out at the autonomy you’ve set.
• 09:12 A counterparty limit breach arrives as a proposed fix, not a finding.
• 16:00 Placement of excess cash prepared for approval before the cut-off.

By the time your team logs in, the day is already run. They review it instead of assembling it.

The half nobody is writing about

The same change that lets you put idle and operating cash to work overnight carries a second consequence. A payment you approve is now final in moments, not days.

That window between decision and settlement used to be where errors were caught. As it closes, controls can’t run after the money moves. They have to run before, on every action, at any hour.

Boards are asking treasury to move faster than governance currently supports. That’s a fair position to be in, and it’s why controls aren’t the brake on always-on treasury. They’re the reason it’s possible.

You decide how much runs on its own

Always-on doesn’t mean hands-free. It means you choose, workflow by workflow, how much runs without a person clicking approve:

• Manual. Every move is prepared for you. Nothing happens until a person approves it.
• Semi-autonomous. Agents act inside your guardrails and report as they go.
• Fully autonomous. Agents run the workflow end to end. Your team checks in when it wants.

Overnight sweeps might run on their own while FX hedging still asks first. A workflow moves up when your own approval record shows it has earned it. Raising autonomy changes who clicks, never what is permitted.

The question that follows: who do you partner with?

Most treasuries will build this capability the practical way, one vendor at a time. A forecasting tool here, a payments provider there, a separate partner for digital rails.

Each addition makes sense on its own. Each one is also another counterparty and another place your audit trail breaks.

Ripple Treasury is built as the alternative. Adaptable solutions you adopt at your own pace, with GSmart agents running across all of them inside your policy, and rail-agnostic connectivity from traditional banking to digital infrastructure when you choose it. The orchestration layer underneath is what makes that one operation instead of five.

GSmart works inside your environment: your data isn’t shared with third parties or used to train models elsewhere, and every output is traceable and auditable.

One partner. One counterparty. One audit trail.

Availability of products can vary and may be provided by different Ripple entities.

Where to go from here

Treasury didn’t speed up. It changed shape. The teams that do well in the next decade will be the ones that set their controls for the clock the market now keeps.

See what a treasury day looks like when it’s already run →

‍