AI in the audit
At a recent AICPA “A&A Focus” webcast, panellists explored how auditors can get more from artificial intelligence – not by upgrading the tech, but by upgrading their prompts. Danielle Supkis Cheek, CPA and AI lead at Caseware, introduced a structured approach for writing AI prompts using the STAR method: Situation, Task, Appearance, Refine.
The method helps auditors guide generative AI through complex scenarios, such as reviewing lease agreements under ASC 842. The session showcased how specific, well-structured prompts can generate more reliable, better formatted outputs. It’s a leap beyond casual experimentation, offering a blueprint for practitioners to look further at what works, and what doesn’t.
While the focus was squarely on unstructured documents and audit tasks, the implications resonate across the financial reporting ecosystem.
As we’ve discussed before (see our recent webinar elsewhere in this newsletter), the more structured and machine-readable the source data, the more powerful and reliable AI outputs can be. Combining structured data inputs with tailored, careful prompts looks like a promising area for AI assistance in audit workflows.
Read more here.
