RSM launches AI-guided internal audit platform "Beacon" with Andera

Top 10 firm RSM, in cooperation with AI-powered audit and compliance technology Andera, announced a new AI-driven control testing platform for internal audit functions called RSM Beacon. Andera's AI technology will be integrated into RSM's next-generation audit platform and use RSM's own proprietary testing methodology. 

Processing Content

RSM Beacon automates the preparation of control test work papers. An auditor provides the control description, testing attributes and supporting evidence; the platform reads the evidence, tests each attribute, and produces a fully annotated, review-ready work paper that mirrors how the team would have prepared it manually. It handles the full range of evidence auditors actually work with, including complex Excel workbooks with embedded files, screenshots and annotations. It covers high-volume, multi-sample controls, not just simple ones. RSM is also evaluating the use of the platform on control frameworks beyond SOX, including SOC, NIST, PCI and operational audits.

It is based on an agent-based AI workflow. The system ingests the control, its attributes and the evidence files, then parses those files the way an auditor would read them, including the nuances of how auditors and accountants use Excel (embedded images, tick marks, initials, colored tabs, etc.). From there it allocates evidence to the correct samples, tests each attribute against the evidence, and documents its reasoning at every step. Every conclusion is traceable, with reviewers seeing the full reasoning log alongside the underlying evidence. Critically, the AI does not make judgment calls. Anything ambiguous is flagged as an open item for a human auditor to review.

When using the platform, an auditor would load a control (manually, from a test sheet or RCM upload, or via integration), optionally upload prior-year work papers and reference documents like policies and narratives to give the AI context, and then drag and drop the evidence files. The AI sorts the evidence to the right samples and the auditor can start testing. Tests run in the background, so auditors can queue multiple controls and return to completed work papers, then move into review, where they examine the AI's conclusions, resolve open items and apply professional judgment. 

Andera's underlying technology also powers Andera's own commercial product, which enterprise audit teams license directly, and is in use at large public companies today. A spokesperson, however, clarified that the two are fully segregated environments: client data processed through Beacon stays within Beacon and is never used to train or improve the commercial product.

The platform is currently live and in use by RSM professionals.


For reprint and licensing requests for this article, click here.
Technology Artificial Intelligence Audit Software integration Audit software
MORE FROM ACCOUNTING TODAY
Load More