A software engineering intern at Ramp describes three accounting automation projects built during a fall internship: an AI-powered amortization detection system that analyzes receipt data to suggest prepaid expense schedules, a native amortization engine that creates and syncs journal entries directly within Ramp (bypassing spreadsheet workarounds for ERPs like QuickBooks), and an automated mapping rules system that mines historical transaction data nightly to suggest or auto-create ERP field mappings. The post emphasizes measuring credibility through detection rates, false positives, and override rates, and highlights the importance of understanding actual accountant workflows before choosing technology.
Table of contents
The throughline: make accounting automation feel invisible1) Amortization Automation: from “receipt review” to “prepaid suggestions”2) Ramp‑Native Amortization: building the engine for ERPs that don’t have one3) Automated Suggested Mapping Rules: automation that creates more automationWhat I learned (and what I’d tell future interns)ReflectionWrapping up6 Impressions