How Snowflake’s Internal Audit Team Cut Fieldwork By 50% with AI

Snowflake’s Internal Audit team has transformed its approach by integrating AI to automate evidence collection, testing, and report production, cutting fieldwork time by 50%. The shift enables continuous monitoring of controls instead of periodic historical checks, allowing the team to identify risks in near real time. By leveraging data already centralized in Snowflake, the team avoids the delays of stitching together disparate sources, accelerating audit cycles and improving accuracy.

The team uses Snowflake CoCo, an AI coding agent, to build automated workflows that scan for anomalies in system access logs, financial transactions, and audit trails. A CoCo-built app validates revenue recognition terms across orders, testing every transaction and flagging discrepancies immediately. This replaces manual sampling, which was slow and error-prone, with full-population testing that catches issues as they occur.

To address payroll reconciliation, a CoCo-developed data app now automates the matching of HR and payroll reports, reducing the manual workload by over 500 hours annually. The team also deployed a "Report-as-a-Service" architecture that ingests audit trails, enabling continuous monitoring and saving 450+ hours in annual testing. These tools free control owners to focus on higher-value tasks rather than routine reconciliations.

SOX documentation drafting, once a weeks-long process, is now completed in hours using Petual.ai, a Snowflake Native App that drafts updated templates for auditor review. The app has analyzed over 20 third-party SOC 1 reports, identifying control exceptions automatically and reducing preparation time by days. This shift allows auditors to prioritize high-risk findings and strategic advisory work over administrative tasks.