The year we missed 14 days of December entries
James McKinnon spent six years in public accounting — three at a regional firm in Boston, three as an internal audit manager at a mid-cap manufacturer. In 2022, his team completed a year-end audit for a client where the controller had posted 14 backdated adjusting entries across two accounts in December. The entries were individually below materiality. Collectively, they shifted $2.3M of inventory impairment into the following fiscal year.
The entries were discovered — by the client's new CFO, not the audit team — four months after the engagement closed. The firm issued a restatement. Two senior managers resigned. The audit committee asked why a tool designed to detect exactly this pattern wasn't in use. There wasn't a good answer.
James left the firm in early 2023 and spent six months building the first version of AuditPulsar. The initial prototype was a Python script that ran Benford's Law analysis against a trial balance export. By mid-2023, it had a gradient-boosted ML model trained on historical journal entry data. By late 2023, it had its first three paying clients — all firms that had experienced their own version of the same story.
Today AuditPulsar serves accounting firms across the US, from sole practitioners to regional practices with 30+ professionals. The core mission hasn't changed: find the entries that shouldn't be there, before the engagement closes.
