LedgerGuard is an AI-powered accounting audit platform built for accounting firms and bookkeepers. It connects natively to a firm's QuickBooks Online (Intuit) data and uses Claude AI to audit every client's books every day — flagging categorization errors, unreconciled items, and missing documentation before they compound. The platform replaces manual, spreadsheet-based month-end reconciliation with a single dashboard covering client management, automated analysis, custom firm rules, and daily review summaries.
Built a native QuickBooks (Intuit) OAuth integration enabling real-time, direct data sync from a firm's general ledger — no CSV exports or manual data migration required.
Developed a Claude AI-powered audit engine that reviews every connected client daily against the firm's configured coding rules, materiality thresholds, and review workflows.
Implemented a rules engine allowing firms to define their own coding conventions and materiality thresholds, or start from pre-built defaults.
Built an automated issue-flagging system that surfaces categorization errors, unreconciled items, and missing documents with line-level detail, amounts, and context on exactly what's wrong and how to fix it.
Designed a client management module giving firms a centralized view across their entire client book instead of per-client spreadsheets.
Built review summary reporting so staff receive fix-it flags, managers get client summaries, and partners get a single consolidated dashboard.
Architected the backend with NestJS and Supabase (PostgreSQL), paired with a Next.js frontend for the firm-facing dashboard.
