First in a three-part series on Project Agorá, which is testing tokenisation to improve cross-border payments. It will be followed by a comparison of this central bank-backed initiative with competing projects. A third article, on the lessons for treasurers, completes the series.
A public-private project led by the Bank for International Settlements (BIS), Agorá aims to make wholesale cross-border payments faster and safer through a shared, programmable platform on which tokenised central bank reserves and commercial bank deposits circulate. In July, real-value tests covered about CHF 800,000 in six currencies, with an average settlement time of about 80 seconds.
At the latest EuroFinance conference in Barcelona, Morten Bech, head of the BIS Innovation Hub Swiss Centre and co-chair of the project, gave an update on the platform’s progress. It remains a prototype, with no launch date.
Agorá
Launched in April 2024 under the auspices of the BIS and the Institute of International Finance (IIF), the latter tasked with bringing commercial banks on board, Agorá seeks to show that a shared, programmable ledger can improve how correspondent banking chains handle cross-border payments: shorter processing times, more transparent payments and lower risk for banks.
The prototype covers three points: atomic settlement across currencies, without breaking the relationship between depositors and their bank; streamlining the pre-checks currently duplicated by several banks in the chain, without changing existing standards; and improving sanctions screening and anti-money laundering and counter-terrorist financing (AML/CFT) controls. The project thus keeps the correspondent banking model, with central banks and bank deposits at its core.
Eight central banks take part, including the Banque de France for the Eurosystem, the Federal Reserve Bank of New York and, since May, the Bank of Canada.
On the private side, more than 40 regulated institutions are involved, including the US banks JPMorgan, Citi, BNY and Wells Fargo, the UK banks HSBC and Barclays, and BNP Paribas, Crédit Agricole CIB and Groupe BPCE for France.
Agorá targets wholesale payments, which account for about 91% of cross-border payment value, including large corporate payments. Retail remittances are out of scope. In practice, commercial banks and central banks record their own liabilities, deposits and reserves, as tokens on the same platform. The arrangement preserves the two-tier system: central banks serve banks, and banks serve their clients.
According to the BIS legal analysis, tokenisation does not change the legal nature of the underlying reserves and deposits. A tokenised deposit remains a deposit with the same bank, exchangeable at par, and not a new crypto-asset.
How it works
In Agorá, tokenised deposits are recorded on a common ledger, known as the unifying ledger. Tokenised reserves are recorded on separate ledgers, one per jurisdiction, so that each central bank keeps control of its currency.
The main change is the order of operations. Today, messages and funds travel together along a chain of correspondent banks. An error sends both instructions and funds back the other way. Agorá takes a different approach. The process first aligns all the information (payee, payment path, compliance check results, FX amount), then locks the funds and settles every leg of the payment at the same time.
Bech used an analogy he had heard elsewhere: a group wedding, where either everyone says yes or nobody gets married. If a check fails or a party cannot lock its funds in time, the payment does not happen: there is nothing to unwind or reconcile.
Agorá payment life cycle · 5 stages, 2 exit points
Tokenised deposits move on the shared unifying ledger, tokenised reserves on each central bank’s own ledger. A payment coordinator executes every leg together or cancels them all.
Each bank still runs its own sanctions, AML/CFT and fraud checks, but in parallel rather than one after another. Only the result, pass or fail, is shared along the chain. Only the banks party to the transaction see its details.
Settlement finality was examined from a legal standpoint. Its definition varies from one country to another. The project’s lawyers consider settlement finality achievable in all seven original jurisdictions, through provisions that would sit in the platform’s rulebooks. At its interfaces, the prototype uses the ISO 20022 standard (pacs.008, pacs.009, camt.053) and legal entity identifiers (LEIs), so banks’ existing systems can connect.
What the tests showed
Phase I met its objectives: a working prototype, tested by participating institutions and backed by legal analysis, described in a 97-page report published on 27 May. In July, real-value tests put actual money through it.
| Real-value tests, July 2026 | Result |
|---|---|
| Participants | 28 banks and central banks in Asia, Europe and North America |
| Volume | About CHF 800,000 in 30 transactions, across 17 scenarios |
| Currencies | CHF, EUR, GBP, JPY, KRW, USD |
| Transaction size | CHF 9,000 to CHF 125,000, or equivalent |
| Use cases | Corporate and interbank payments in one or two currencies, payment-versus-payment (PvP), intragroup transfers |
| Settlement time | About 80 seconds on average from initiation to settlement, without integration with real-time gross settlement (RTGS) systems or core banking |
| Operators | SIX as operational facilitator, Kaleido as technology provider |
Participants valued being able to track payment status and routing end to end. According to Bech, two of the test transactions involved corporate treasurers, who saw the potential.
The report also sets out the limits. The prototype was not designed for the volumes, resilience or cyber security of a production service. It does not include liquidity-saving mechanisms or FX execution. During the tests, the cross-currency amount was entered manually. Cases requiring a compliance investigation can still interrupt straight-through processing (STP). Finally, who would own and operate the shared ledger remains an open question.
Nor is cost reduction among the priority objectives. For the project’s sponsors, it depends on commercial rather than technical choices.
Next steps
Agorá is now moving from feasibility to viability. The project’s co-chair identified two difficult questions: the business model and the governance of a global infrastructure spanning many jurisdictions.
Further real-value tests are planned by the end of 2026, including in Canadian dollars, according to Bech. The BIS expects the private sector to play a greater role in the next phase, with central banks still closely involved. The report sets out further work on liquidity-saving mechanisms, coordinated AML/CFT and sanctions controls, cyber resilience, interoperability and governance rules.
Bech gave no date. He cited CLS as an example of infrastructure that took decades to build, adding that Agorá does not intend to follow that path. He sees the current period as a window to decide whether the project should go further.
Sources
- Project Agorá: tokenisation in practice, presentation and Q&A with Morten Bech, EuroFinance, Barcelona, 18 September 2026
- Project Agorá project page and real-value test results, BIS
- Project Agorá: a shared programmable platform for wholesale cross-border payments, BIS Innovation Hub report, 27 May 2026
- Press release on the prototype report, BIS, 27 May 2026
- Project Agorá FAQ, BIS
- New participants join Project Agorá, Ledger Insights, 29 September 2026