Company overview

Company name: YH Digital Bookkeeping
Industry: Accounting
Location: Singapore
Company size: Three-person accounting firm
Products used: fileLedger
Use Cases: E2E Client Bookkeeping Tasks

Growing client demand with a fixed team and a manual document process

For a small accounting practice, the capacity ceiling is rarely about skill or ambition. It is almost always about process. Tan Yee Hou, Founder of a three-person firm in Singapore, understood this clearly. His team was technically capable and commercially motivated. What constrained them was the volume and variety of manual document work that sat between each client engagement and a completed set of accounts.

Multi-line invoices required field-by-field manual entry. Vendor bills needed to be sorted, cross-referenced against Xero, and individually attached to the relevant records. Invoice references were copy-pasted into spreadsheets by hand before reconciliation could begin. Each of these tasks, considered in isolation, appeared manageable. Considered together, they formed a process that was slow, error-prone, and structurally unable to scale.

The risk was not simply one of efficiency. Manual data entry at volume creates conditions where errors are not the exception; they are an expected cost of doing business. A distraction mid-task, an incorrectly keyed reference, a misapplied GST rate, any one of these could require the entire entry to be redone. For a team managing multiple clients concurrently, that cost accumulated quickly.

Taking on additional clients without resolving the underlying workflow would mean more of the same: more manual entry, more error exposure, more pressure on a team already operating at capacity. Growth required a different approach to document processing, not more hours.

Removing document handling from the daily workflow entirely

The solution Tan Yee Hou implemented with fileAI is built around a straightforward principle: documents should flow into the system automatically, be processed without manual intervention, and surface only the exceptions that genuinely require human judgement. Everything else should run on its own.

Continuous document ingestion

Clients are assigned a dedicated email alias. When vendors copy that address on invoices, the documents flow directly into fileAI via a Gmail integration. There is no manual retrieval, no chasing of missing paperwork, and no year-end reconciliation of what arrived and what did not. The firm estimates this has reduced document-related client queries by 80 to 90 percent.

For year-end catch-ups, clients drop their files into a shared folder. The accountant downloads the folder in full and uploads it to fileAI as a single batch. Duplicate documents are detected and discarded automatically, with no manual review required and no charge incurred for them.

“You just download everything and throw it up. That is how I do the work.”

Tan Yee Hou

Founder

Rules-based processing for recurring suppliers

For vendors the firm processes regularly, fileAI applies configured rules on each subsequent upload: the correct account codes, the applicable GST rate, and the handling logic specific to that supplier’s invoice format. Senior accountants set these rules once. Junior staff follow the workflow without needing to interpret or second-guess each document. Exceptions are surfaced for review automatically rather than being absorbed silently into the output.

The practical effect is that document ingestion has moved from an ongoing source of error and delay into a governed, repeatable process. It is infrastructure rather than administration.

Measurable gains across capacity, accuracy, and service scope

The impact of the fileAI implementation has been consistent across every dimension the firm tracks: throughput, accuracy, client service, and commercial scope.

Xero had already multiplied the firm’s output relative to a fully manual operation. fileAI added a further 300 percent efficiency gain on top of that, bringing the combined effect to the equivalent output of approximately fifteen staff members. The gain is not simply a function of volume. It is also a function of consistency: the same rules applied to every document, every time, without the variability that manual processing introduces.

The seven most active client accounts, which previously required the better part of a month to reconcile and report, are now complete within the first working week. The capacity that has been freed is being reinvested in client relationships and higher-value advisory work, rather than absorbed by additional headcount.

“The quality of my work will drop. Or I would have to spend a lot more time to get the quality that I am getting.”

Tan Yee Hou

Founder, on what would happen if fileAI were removed

The compounding value of removing friction at every step

The most important insight from Tan Yee Hou’s experience is not about any individual feature. It is about what happens when every source of manual friction across a document-heavy workflow is removed simultaneously.