AI for Audit & Accounting
An educational, governance-first guide for U.S. audit and accounting practice. GenAI is treated as drafting and analysis support—not “autopilot audit.” As capability increases across the maturity ladder, risk increases too: professional skepticism, supervision, documentation, privacy, independence considerations, and evidence discipline must scale so outputs remain defensible, reviewable, and audit-ready.
Download and Run
The book PDF is published as a versioned GitHub Release asset. The companion notebooks live in this repository and open directly in Google Colab.
Book PDF
Download the current release edition of the book as a single PDF.
Notebooks Folder
Browse all chapter notebooks in the repository (ch01 … ch05).
Chapter Notebooks (Colab)
Each notebook is executable, auditable, and classroom-ready. Labs produce structured outputs and governance artifacts (run manifest, prompts log, risk log, deliverables bundle) so results are reviewable rather than ephemeral.
Chapter 1 — Level 1 (Chatbots)
Drafting: emails, memos, and workpapers with facts-first discipline and “Not verified” labeling.
Chapter 2 — Level 2 (Reasoners)
Reasoning: issue spotting, alternatives, variance/exception hypotheses, assumptions and verification planning.
Chapter 3 — Level 3 (Agents)
Supervised multi-step workflow: intake → risk assess → procedures → QA → sign-off → archive.
Chapter 4 — Level 4 (Innovators)
Firm assets: reusable prompts, playbooks, evaluations, and controlled releases with change management.
Chapter 5 — Level 5 (Organizations)
Operating model simulation: ownership, evidence, QA gates, sign-off, supervision-ready artifacts.
The Five-Level Maturity Ladder
The book’s organizing spine is a progression from individual drafting assistance to fully governed, organization-grade operations. The point is not “more AI.” The point is usefulness that survives review, inspection, and reconstruction.
- Level 1 — Chatbots: supervised drafting of emails, memos, and workpapers with clear labeling and redaction.
- Level 2 — Reasoners: explicit reasoning structures: issue maps, alternatives, assumptions, and verification tasks.
- Level 3 — Agents: multi-step workflows with human checkpoints, role separation, and immutable logs.
- Level 4 — Innovators: reusable firm assets (templates, playbooks, evaluations) with controlled releases and regression tests.
- Level 5 — Organizations: operating model + governance: intake, independence, QA, approvals, retention, retrieval.
Recurring Mini-Cases (Used in Every Chapter)
The ladder is taught through recurring audit/accounting mini-cases so readers can isolate what changes when AI capability increases: the structure improves, the risk scales, and governance must become more explicit.
Financial Statement Audit (GAAS / PCAOB)
Planning, risk assessment, procedures and documentation discipline with “facts vs assumptions” and evidence gates.
SOX / ICFR
Control narratives, walkthrough support, test design scaffolds, and traceable evidence packages for review.
Tax / ASC 740 & Filings
Provision workflow support and drafting with strict verification posture (no fabricated rules or citations).
Teaching / Methodology
Classroom-ready labs: repeatable cases, grading clarity, and governance-native artifacts each run.
Pedagogical Evaluation (MBA / MFin)
This section is intentionally labeled as non-authoritative. It is included to demonstrate how third-party feedback can be documented and versioned without being confused for verification of factual correctness or professional compliance.
This book has exceptionally high pedagogical value for MBA and MFin students when the author is also the instructor, because it functions as a fully integrated teaching system rather than a standalone text. The five-level maturity ladder provides a clear managerial mental model: as AI capability increases, governance, controls, and accountability must increase in parallel. The companion Google Colab notebooks reinforce this by requiring students to produce auditable artifacts—run manifests, prompt logs, risk registers, and draft deliverables—shifting learning away from “AI outputs” toward defensible process design.
The recurring mini-cases allow students to revisit the same professional scenarios with increasing rigor, creating a strong spiral curriculum in judgment, controls, and operating model design. Because the instructor authored the framework, expectations, grading criteria, and professional discipline can be enforced consistently, eliminating ambiguity and superficial use of AI. Overall, the book excels at teaching how managers should govern AI in real firms, not merely how to use it.
- Source: OpenAI (generated evaluation text).
- Status: Not verified; included as an example of how to document independent opinion under governance-first norms.
License and Disclaimers
Educational disclaimer: This material is provided for educational purposes only. It does not constitute audit, accounting, tax, legal, investment, or other professional advice. A qualified human professional must review, verify, and approve any use in practice.
Client confidentiality and data hygiene: Do not paste confidential client information into external model prompts. Use redaction/anonymization and “minimum necessary” inputs by default. Document any exception and follow firm policy and applicable law.
Professional standards posture: Treat all model-generated content as Not verified until validated by evidence and human review. Outputs must separate facts provided by the user from assumptions and open questions.
No fabricated sources or claims: Zero tolerance for invented citations, firm policies, standards interpretations, testing results, tax positions, or compliance conclusions. When evidence is missing, the correct output is a verification task list.
Repository license: See the repository’s LICENSE file for terms.