For decades, the accounting process has been bottlenecked by the same problem: getting transaction data out of banks and into accounting systems. Open banking is eliminating that bottleneck, and the implications for accounting automation are profound.
What Open Banking Actually Is
Open banking is a regulatory and technical framework that requires banks to share customer account data with authorised third parties through secure APIs, with the customer's explicit consent. In Europe, this is mandated by the revised Payment Services Directive (PSD2), which came into effect in January 2018 and was reinforced by the Regulatory Technical Standards on Strong Customer Authentication.
The European Commission proposed PSD3 and the Payment Services Regulation (PSR) in June 2023, signalling an evolution toward even broader data sharing. The proposed Financial Data Access (FIDA) framework would extend open banking principles beyond payment accounts to include savings, investments, insurance, and pensions.
In practical terms, open banking means a business can authorise its accounting software to connect directly to its bank account and retrieve transaction data through a standardised API. No more downloading CSV files. No more waiting for monthly statements. No more manual data entry of bank transactions.
From Monthly Statements to Real-Time Data
The traditional accounting cycle revolves around periodic data: monthly bank statements, quarterly VAT returns, annual financial statements. This periodicity is not a feature of accounting theory. It is a limitation of data availability. Accountants work in batches because that is how they receive information.
Open banking changes the fundamental timing of accounting data. Instead of reconciling transactions against a month-end bank statement, transactions flow into the accounting system in near real-time. Some open banking APIs provide transaction data within minutes of the payment being processed.
This shift from batch to stream has cascading effects:
- Cash flow visibility moves from a monthly snapshot to a continuous view
- Reconciliation can happen transaction-by-transaction as each one arrives, rather than in a monthly batch
- Anomaly detection can flag unusual transactions within hours, not weeks after the fact
- VAT calculations can be updated continuously, eliminating the quarter-end scramble
Automatic Reconciliation
Bank reconciliation, matching bank transactions to accounting entries, is one of the most time-consuming tasks in bookkeeping. For a small business with 200 transactions per month, manual reconciliation might take an accountant two to three hours. For a busier operation, significantly more.
Open banking data comes with structured metadata that manual bank statements lack. API responses typically include merchant category codes, payment references, counterparty details, and timestamps accurate to the second. This structured data makes automatic matching dramatically more reliable.
When a payment of EUR 1,247.50 appears in the bank feed with a reference matching invoice INV-2024-0847, the system can match these automatically with near certainty. When combined with AI-driven pattern recognition, even transactions without clear references can be matched. The system learns that the monthly payment of approximately EUR 340 to "VODAFONE MT" corresponds to the telecoms expense category and the recurring Vodafone invoice.
Xero reported that its automatic bank reconciliation feature handles over 75% of transactions without human intervention for established businesses. The remaining 25% typically involve new suppliers, unusual transactions, or ambiguous descriptions that benefit from human judgment.
Where Malta Stands
Malta's open banking landscape is still developing. The Malta Financial Services Authority (MFSA) regulates payment institutions and electronic money institutions under PSD2, and several licensed Account Information Service Providers (AISPs) and Payment Initiation Service Providers (PISPs) operate in the market.
However, the practical availability of open banking connections to Maltese bank accounts has lagged behind markets like the UK, where the Competition and Markets Authority's Open Banking Implementation Entity drove rapid adoption. Bank of Valletta and HSBC Malta offer API access in compliance with PSD2, but the ecosystem of third-party applications leveraging these APIs is smaller than in larger European markets.
The European Commission's push toward FIDA and PSD3 should accelerate adoption across smaller EU member states. For Maltese businesses, the trajectory is clear even if the timeline is slower: direct API connections between banks and accounting systems will replace manual data transfers.
The VAT Implications
For VAT-registered businesses, open banking creates the possibility of a fundamentally different compliance workflow. Consider the current process: a business receives an invoice, pays it, records the transaction, categorises it, determines the VAT treatment, and eventually includes it in a quarterly return. Each step involves potential delay and error.
With open banking and AI working together, the process compresses. The payment appears in the accounting system via the bank API. The AI matches it to the corresponding invoice (also digitally captured). The VAT treatment is determined automatically based on the supplier, the goods or services, and the applicable rules. The transaction is recorded, categorised, and reflected in the running VAT position immediately.
This matters because VAT errors compound. A miscategorised transaction in January might not be caught until the Q1 return is prepared in April. By then, the original receipt may be difficult to find, the context is forgotten, and the error is either perpetuated or requires disproportionate effort to correct. Real-time processing catches errors when context is fresh and correction is simple.
AI Making Sense of the Data
Open banking provides the data. AI provides the intelligence to interpret it. Raw bank transaction data is surprisingly messy. Merchant names are abbreviated, truncated, or inconsistent. Payment references may be missing or formatted differently by different banks. The same supplier might appear under multiple merchant identifiers across different payment methods.
AI-driven categorisation models learn these patterns. They recognise that "MSFTMSBILL" and "MICROSOFTOFFICE" and "MICROSOFT 365" all represent the same subscription service. They learn that a business in the real estate sector making frequent payments to the Malta Tourism Authority is likely paying licensing fees, not booking holidays.
The accuracy of these models improves with volume. A system that has processed millions of business transactions has seen enough variations of merchant names, payment patterns, and expense categories to handle edge cases that would stump a rule-based system. Published benchmarks from accounting software providers suggest categorisation accuracy rates of 85-92% on first attempt, rising to 95%+ after a few months of use as the model learns business-specific patterns.
Where This Is Heading
The logical endpoint of open banking combined with AI is real-time accounting. Not accounting that happens faster. Accounting that happens continuously, automatically, and correctly, without the concept of "closing the books" at the end of a period.
Imagine a system where every transaction is categorised, reconciled, and reflected in financial statements the moment it occurs. Where the VAT position is always current, not reconstructed quarterly. Where cash flow forecasts update with every payment received and made. Where the annual financial statements are, in effect, always complete and waiting for the final transaction of the year.
This is not science fiction. The component technologies exist today. Open banking provides the data stream. AI provides the intelligence. Cloud computing provides the infrastructure. What remains is integration, refinement, and the inevitable catching-up of regulation with technological capability.
The transition will not be instantaneous. Legacy banking systems, varying API quality across institutions, and the inherent complexity of tax law across jurisdictions all impose friction. But the direction is unambiguous. The manual data entry, batch reconciliation, and periodic reporting that define traditional accounting are artefacts of technological limitations that are being systematically removed.
Michael Cutajar, CPA — Founder of Accora.