Case study — 2024 — 2026 · internally "Perfecto"
Tallify
AI bookkeeping for insurance agencies. An AI-assisted bookkeeping and carrier-commission reconciliation platform built for independent US insurance agencies — bank feeds in, GAAP-compliant books and reconciled commission statements out.
- Type
- saas
- Years
- 2024 — 2026
- Stack
- Next.js · Node.js · Python · PostgreSQL · Plaid · Azure OCR · OpenAI API · Redis
The problem
Independent agencies reconcile commission statements from a dozen carriers against their books by hand. It leaks money, eats nights, and turns tax season into a scramble.
My role
Core engineer at Xassure by eDesk: bank integration layer, AI categorization pipeline, OCR ingestion, and the reporting dashboard.
The hard part
Financial data allows zero tolerance for silent errors. The pipeline links banks via Plaid, scans receipts and invoices with OCR, categorizes transactions with AI, and still has to keep every number auditable — AI suggestions are tracked, human-reviewable, and reversible rather than silently applied. The system processes 50K+ transactions a month.
Outcome
Live at tallify.ai serving independent insurance agencies, with plans from $349/mo, integrations across QuickBooks, Xero, and agency management systems, plus a conversational AI interface for financial insights and forecasting.
