Every tax season, we hear the same question from Malaysian SME owners: “How do I produce my business accounts?” In this era of digitalisation, many small SMEs already sell online, collect digital payments, and operate real trading businesses. Yet, from a finance operations perspective, the irony is clear: many still lack a formal accounting system. In practice, their most reliable business record is often their bank statement.
The standard response has always been the same: “You need an accounting system.” That answer is technically correct, but it does not address the operational reality. Many small businesses have been trading for years without a formal system, and their teams remain fully occupied with selling, purchasing stock, paying suppliers, processing refunds, managing cash flow, and resolving daily operational issues.
When the advice never changes, the outcome rarely changes either. A meaningful number of SMEs still do not file returns at all, not because they intend to be non-compliant, but because the clerical workload is too heavy. They are not only wrestling with paperwork; they are managing survival, cash flow, sales, operations, staff, platforms, suppliers, and the daily pressure of running a business.
From the finance and accounting side at AINNA, I see AI playing a practical role for SMEs, not as a magic button, not as a replacement for accountants, and not as a shortcut to bypass compliance. AI should function as a reconciliation bridge between informal, bank-statement-based records and structured, reviewable, accountant-ready financial data.
In one of our internal finance operations initiatives, we explored how AI can support finance preparation using a detached-system approach:
Detached system: each finance area runs as a separate workflow, such as sales, refunds, purchases, operating expenses, loans, fixed assets, inventory, owner movements, and internal transfers.
Segmentation: transactions are classified based on economic substance and chart-of-accounts treatment, not only on bank descriptions.
Guardrails: predefined rules prevent the AI from guessing, especially for loan principal, refunds, fixed deposits, internal transfers, and unclear transactions.
Smart routing: each transaction is directed to the correct place, the Profit & Loss statement, Balance Sheet, supporting schedule, or human review list.
The important point from a cost-control standpoint is this: an AI workflow built this way does not always require premium models, expensive infrastructure, or a system that runs around the clock. Because the workflows are detached, processing can be event-driven, and AI models are invoked only for specific tasks. That makes the approach more cost-efficient, more controlled, and more realistic for small SMEs.
With this approach, more small SMEs may finally have a realistic pathway to declare their business properly, especially if structured cash-basis reports prepared from bank statements can be accepted as a starting point, subject to proper review. Compliance frameworks should not only be designed for SMEs that already have perfect systems. They must also create a practical pathway for SMEs that are still starting from bank statements. For me, this is the real opportunity: using AI-assisted detached systems to help SMEs become more organized, more compliant, and more confident in their financial position.
