Finance That Never Closes: Why Businesses Need to Prepare Now

It is still easy to see tokenisation as someone else's problem. A company has customers, accounting software, a bank account and an established process for approving payments. Artificial intelligence helps write a proposal or summarise a contract. The business is running. Why open another technological chapter?

Because the environment in which it operates is gradually changing. New technical possibilities are emerging in the relationship between business decisions, money, ownership and control. This is where differences in companies' speed, costs and trustworthiness may develop. They can remain barely visible from the outside for a long time. Within a business relationship, however, they determine who can substantiate an order's status within minutes, who manages to release collateral and who spends another day looking for the right attachment.

The news we followed over the past week gives this development a concrete shape. Some events took place between 26 October and 1 November; others were covered during that period in our articles following announcements from the previous week. Together, they show where infrastructure is being built today and where workable business models are still taking shape.

My central argument is simple: waiting until everything is ready before preparing means leaving the first lessons to competitors. Preparation begins much closer to your own business than the grand technological slogans suggest.

2028 is an infrastructure milestone. Preparation runs on a different clock

On 29 October, ECB Executive Board member Piero Cipollone outlined the next stages of Pontes: significantly longer operating hours by mid-2027 and availability 24 hours a day, seven days a week, throughout the year in 2028. This is the Eurosystem's announced plan. Source: ECB.

To the owner of a manufacturing business, that announcement may seem remote. To someone responsible for liquidity, financing or investment settlement, it has very practical implications. If a transaction can take place outside normal operating hours, authorisations, available funds, exception handling and the ability to establish its actual outcome must all be ready to support it. The operating window of financial infrastructure and the operational readiness of an individual company are two separate matters.

We already know continuous payments in some areas. Extending those capabilities to tokenised assets introduces further relationships: who holds the asset, under what conditions it can be transferred and when payment is truly final. A system available on a Saturday evening will not help a company whose decisive approval authority is trapped in a Monday email.

I therefore read 2028 as a reference point for investment and preparation. It makes sense to begin identifying which processes a company needs to change, which providers it can connect with and where human decisions remain necessary. The ECB's announcement does not, however, establish that every agent service will automatically be ready that year or that every company will gain direct access to central bank money.

The motorway is gaining real traffic

On 26 October, the European Stability Mechanism (ESM) announced the pricing of a €1 billion digital bond issue intended to settle through Pontes. As of this commentary's publication, we are relying on the announced settlement plan; we are not describing settlement as completed. Source: Clearstream.

The size of the transaction explains why institutions are worth watching. Their infrastructure is where legal rules, technology, cash settlement and accountability are being brought together in transactions with real economic weight. The size of the issue alone does not demonstrate that every smaller business can use the same route tomorrow.

The motorway analogy is useful. A main route needs capacity, safety and shared rules. For a local business, it becomes valuable only when there is a usable access road. In digital finance, those connections will take the form of specific services from banks, administrators, technology suppliers and other authorised providers. Their offerings will need a price, a contract, an interface, customer support and a way to deal with things going wrong.

In my view, much of the commercial competition will take place around those connections. A provider that translates an institutional capability into a service a business can understand may build a lasting customer relationship. A company that first puts its data and authority in order will be better placed to compare such offerings. The motorway itself will not fix a disorganised warehouse or an invalid power of attorney.

One network does not mean one legal world

In October, J.P. Morgan extended JPM Coin to Ethereum for approved institutional clients. Malaysia launched SSDS+, connected to RENTAS+ and central bank money. We examined both cases in articles published over the past week. J.P. Morgan, Bank Negara Malaysia.

The shared direction matters: digital asset records are moving closer to the means used to pay for those assets. This creates opportunities to reduce repeated data entry and reconciliation. The presence of money and an asset on the same network does not, by itself, ensure their simultaneous and legally effective exchange. Delivery versus payment requires the entire mechanism to be designed correctly, including failure scenarios and participants' responsibilities.

In such an arrangement, Ethereum is the technological environment. Access to a particular banking product, the right to hold a security, custody rules and permission to trade must be considered separately. Assuming that a public blockchain automatically opens every instrument to every wallet and exchange can lead to costly mistakes.

The different meanings of the word token also need to be distinguished. A token used by a language model to process text is different from a token representing a financial or other right. A payment token that replaces sensitive card details does not necessarily represent an investment asset either. A shared word does not imply shared economic substance. Meaningful cooperation depends on a precise description of rights, money and operating rules.

The Bank for International Settlements has long described the potential of shared programmable environments for money and assets. Such an environment does not always have to be based on blockchain. For business owners, it is therefore more useful to ask which operation a service will simplify and how it will substantiate the outcome than to begin by choosing a network. Source: BIS.

Regulation is moving towards specific conditions of use

Boerse Stuttgart Group obtained a DLT TSS licence for the Seturion platform. Generali launched a tokenised fund class for authorised companies within its own group. The first development expands regulated infrastructure; the second tests investment servicing within a limited group of participants. Their significance lies precisely in their defined scope. Boerse Stuttgart, Generali.

On 26 October, Hong Kong added continuous secondary trading in two tokenised ChinaAMC funds. Three days later, the Bank of England opened a consultation on tokenised collateral at central counterparties. Secondary trading is not the same as continuous subscriptions and redemptions; a consultation is not blanket approval. SFC, Bank of England.

Taken together, these developments show different parts of financial operations evolving in parallel. Issuance, trading, fund administration, collateral and settlement each serve a different purpose. For a business, it will be important to understand which function it actually needs and who is authorised to provide it. A precise question to a supplier may be worth more than a lengthy presentation about the future.

This also creates possibilities for new financing structures. One can imagine a bond with a previously agreed change in its interest rate when a verifiable condition is met. Programmable infrastructure can help calculate and execute that change. The economic entitlement must nevertheless follow from the bond's terms and the data used. Who confirms that the condition has been met? How is an error corrected? What happens in a dispute? Only the answers turn a technological idea into a product. Automation alone does not guarantee cheaper capital.

AI is already inside. What matters is its authority

The Bank of England and FCA survey published on 26 October reports AI use among 86% of respondents. Around half of those answering the detailed questionnaire report ongoing controls for data risks. These are findings among financial firms, not a measure of daily use by all finance professionals and regulators. Source: BoE/FCA.

For company leaders, this calls for a more precise question than “Do we use AI?” What can it access? May it only make recommendations, or may it also act? Who set its limits? Where can we trace what it did and the evidence on which it relied? The distinction between an assistant that helps write text and a system that can spend company money is a distinction in responsibility.

In October, Mastercard demonstrated Agent Pay for Machines and opened early access. The agent makes purchases within the permissions and budget set by the business. Visa, BILL and Billtrust also announced a pilot in which agents execute previously approved invoice payments through supplier portals. Mastercard, Visa.

This is where institutional developments meet an ordinary working day. An order, an approved invoice, a specified supplier and confirmation of payment are familiar to even a small company. An agent can take over some tasks between systems, provided it has suitable interfaces and a precisely limited mandate. It might prepare payment documentation or monitor the outcome of an authorised operation. Supplier selection, the decision to enter into an obligation and payment execution can remain divided among different roles.

The idea of a digital buyer or administrator is acquiring a concrete meaning. Quality will be measured by whether it carried out permitted work, recognised exceptions and left evidence of its intervention. A confident answer in a chat is not that evidence. Operational value becomes apparent in the accounting records, in a complaint or when investigating an unauthorised instruction.

What this could change in an ordinary business

Consider a hypothetical company purchasing materials. It has an approved order in one system, a goods received note in another and an invoice in an email. Today, an employee compares the information, resolves discrepancies and submits the payment for approval. Delays often arise while finding supporting records or explaining an exception, rather than from pressing the payment button itself.

A controlled agent could, in this example, match the order to the delivery and flag an amount that does not agree. If confirmation of receipt is missing, it must not invent it. If the supplier has changed its bank account number, the prescribed verification must follow. If the mandate has expired, the system must stop the action. And if the payment provider has not returned an outcome, the company cannot simply send the same payment again and hope it will not be duplicated.

The benefits of such an arrangement can be measured. How long does reconciliation take? How many payments require manual intervention? How many errors are caught before submission? How quickly can the company establish the actual status? These questions translate technology into operational economics. They also help expose a project that merely adds costs and another screen.

A tokenised asset does not need to be part of every process. It becomes useful where it helps record and transfer a particular right or connect the actions of several participants. A company should be able to explain why a good database and a standard payment are insufficient for its particular purpose. That discipline protects the budget and makes genuinely useful services easier to identify.

Continuous checking depends on knowing whom to trust

The move from reconstructing events after the fact to having evidence continuously available is one of the most interesting possibilities of this transformation. Accounting does not disappear. It remains necessary to recognise rights and obligations, value them and produce financial reports. It can, however, receive more accurate and timely supporting information, with clear provenance and a history of changes.

The analogy with engine management captures the intended benefit well: more information about the current state allows an earlier response. A warning light nevertheless needs a reliable sensor and a correct interpretation of its reading. In a company, that sensor might be confirmation of delivery, a balance or an authorised decision. A well-connected system can spread a faulty input faster than the old process did.

A blockchain validator assesses a transaction against the rules of its network. That does not automatically constitute a financial audit or confirm that goods actually arrived or that a valuation reflects reality. Similarly, a cryptographic hash helps demonstrate the integrity of a record, but it does not make the original claim true. Legal, monetary, technical and accounting outcomes can depend on different authorities.

This is precisely why independent checks on agents are worth developing. One system can verify another's outputs, provided the exercise does not become mutual confirmation of the same error. We need distinct evidence, separated authority, testing and a person or organisation accountable for intervention. Critical operations also require the ability to stop activity in time.

Research covered in our briefing supports this caution. One study examined how deliberately strengthened agent preferences influenced experts; another used a laboratory setting to test the evasion of monitoring by deliberately modified models. This does not demonstrate that deployed AI routinely behaves in this way. Study of judgment calls, laboratory research.

There may also be room for specialised assessments of agent services. A credible rating would require a published methodology, independence, a way to manage conflicts of interest and an opportunity to challenge the result. Automated systems assigning scores to one another will not, by itself, create trust. This should be seen as a direction for developing oversight services, rather than an established market standard.

Speed makes properly defined permissions more valuable

An analysis of the Binance incident on 29 October highlighted the misuse of a signed delegation. The details remain under investigation. For our purposes, the key point is that security also depends on the scope of authority once it has been granted. Source: Fireblocks.

A company needs to know who can change payment recipients, who approves limits and who can revoke a delegation. It must also distinguish a technically valid signature from an authorised business intention. A poorly defined approval may permit an action its originator never intended. In continuous operations, such a detail becomes a fundamental part of risk management.

We also examined Harari's warning about financial instruments whose complexity people may cease to understand in our analysis “AI Will Begin Taking Over the Financial System”. Connecting that warning to tokenisation is our own editorial interpretation. The practical question for company leaders is straightforward: can we explain what happened, who authorised it and how to put it right?

Bureaucracy contains rules, which creates many opportunities for automation. It also contains exceptions, unclear evidence and disputes over what those rules mean. A good service must be able to recognise the boundary of its mandate. That will distinguish it from an impressive demonstration that runs through a prepared example without a single hesitation.

What company leaders can do on Monday

The first step is to choose one specific process. It might be purchasing and paying for materials, managing collateral or preparing documentation for financing. Describe how it works today, its delays, the cost of manual work and its most common errors. Without a baseline measurement, there is nothing against which to compare a supplier's promise.

The second step is to identify authoritative sources. Where is a binding order created? Who confirms delivery? Which system determines the payment's status? Where does accounting obtain its evidence? If different departments give different answers, automation will initially multiply their disagreement.

The third step is to separate authority. Who may propose, approve, execute and check? How long does a permission remain valid, and how is it revoked? Which operations require a second person? These questions are valuable today, even if the company does not connect a single agent for another year.

The fourth step is a small, measurable pilot. Start with test data, then move to a proportionate and controlled scope. It must also cover an incorrect invoice, a changed account, a repeated instruction, a provider outage and an unknown outcome. Handling an exception often reveals more about a system's quality than the speed of an ideal transaction.

The fifth step is to talk to providers in precise terms. Ask about the specific service, legal role, available interface, responsibilities, data portability and total costs. Have them show the audit trail and the procedure for restoring operations. Require evidence of actual availability, rather than a presentation of future prospects. This is how a company learns the language in which digital capital will actually be discussed.

An advantage develops before it becomes visible

The next two years could significantly influence the distribution of experience, customer relationships and supplier positions. That is a strategic judgment, not a certainty that the market will close on a particular day or that an unprepared company will automatically fail. An advantage nevertheless accumulates from things that are difficult to catch up on all at once: good data, tested processes, partners' trust and people who know what to ask.

A business that begins only when a bank, customer or competitor forces it to change will have less time to choose and negotiate. It may pay for haste and accept terms it does not yet understand. A company taking small, verifiable steps today, by contrast, is building its ability to make informed decisions.

The transformation illustrated by this week's developments is already under way. It moves at different speeds, spans different jurisdictions and leaves many questions open. That is precisely why it deserves sustained attention. The purpose of preparation is to understand where new infrastructure can help your business and develop the ability to use it safely.

You can wait for 2028 on a calendar. In business strategy, the work starts today.

---

This editorial commentary draws on the author's dictation and sources checked as of 1 November 2026. Its strategic conclusions and hypothetical business example are the author's analysis. The projects discussed differ in scope and stage of availability.