ISCAP Proceedings: Abstract Presentation
Teaching Agentic AI for Financial Statement Analytics: An EDGAR-Based Case
Lu Lu
Western Illinois University
Rong Zheng
Department of Economics, Applied Statistics, and International Business
Abstract
The use of generative and agentic artificial intelligence in the accounting profession has grown rapidly, and accounting accreditors and employers increasingly call for graduates who can direct AI tools to retrieve, structure, and analyze real financial data rather than questions and chats. This case introduces students, who need no prior programming experience, to direct OpenAI Codex to complete two linked tasks using EDGAR filings. In Part 1, students prompt Codex to extract non-financial data from two competing manufacturers and verify the results against the source documents. In Part 2, students direct Codex to build a repeatable workflow that pulls financial-statement data and computes liquidity, profitability, and leverage ratios for both companies, then interpret the resulting comparison. Students conclude by packaging their prompts and code as a reusable Skill that can be applied to other companies. This case is designed for introductory accounting information systems courses but can be adapted for financial statement analysis, auditing, or any course in which students analyze SEC filings.