AI for Accounting Workpapers: Drafts, Evidence, and Review

AI can help prepare workpaper structures, organize non-sensitive source references, and draft standard documentation sections, but it cannot provide evidence, reach audit conclusions, or replace the reviewer sign-off that gives workpapers their professional significance. Tools like fenbs help track open review questions, source references, and handoffs while the workpapers themselves stay in the approved system.

Accounting workpapers, whether for audit, tax, or internal review, serve a specific purpose: they document the evidence gathered, the procedures performed, and the conclusions reached. Their value comes from the fact that a qualified person reviewed and stood behind them. AI can accelerate the preparation; it cannot substitute for that accountability.

What workpapers require and where AI can help

Workpaper preparation typically involves three layers: gathering evidence, documenting procedures, and reaching a conclusion. AI can contribute to the documentation layer when the accountant has already done the thinking.

• Standard section drafts: producing the structure and headings of a workpaper from a template description, ready for the accountant to fill in with actual evidence.
• Reference organization: listing and linking the documents, standards, and source materials cited in the workpaper.
• Checklist generation: producing a procedure checklist based on the workpaper's objective, which the accountant then reviews and adapts.
• Summary sections: drafting the narrative summary of a completed section from the evidence and procedures the accountant describes.
• Cross-reference checking: identifying where references in the workpaper are inconsistent or where a cited document number does not match the description.

These are productivity contributions, not judgment contributions. The accountant provides the substance; AI organizes the structure.

Evidence standards AI cannot satisfy

Workpaper evidence must be sufficient, appropriate, and obtained through procedures the accountant actually performed. AI cannot observe a physical inventory, confirm a bank balance, inspect a contract, or interview management. It can only process information the accountant provides. Every evidence conclusion in the workpaper must describe something the accountant actually did, not something the AI described as the typical procedure.

The specific risk is fabrication. If an accountant asks AI to "complete" a workpaper section without providing the actual evidence, the result will be plausible-sounding documentation without underlying substance. Workpapers built on AI-generated fabrication create fraud risk, not just error risk. This is a firm line: AI drafts the structure, but the accountant populates it with real evidence.

Reviewer sign-off and professional accountability

Workpapers exist in part to support a reviewer's sign-off. The reviewer assesses whether the procedures performed are sufficient for the objective, whether the evidence supports the conclusion, and whether the documentation is complete. This review cannot be replaced by AI output or by the preparer's own assessment.

PCAOB and AICPA auditing standards specify documentation requirements for audit workpapers. Using AI in workpaper preparation does not alter those requirements. The documentation still needs to show what was done, by whom, when, and what it supported. AI-assisted drafts must meet the same standard as handwritten ones.

Keeping source references and review questions outside the workpaper

Some workpaper-adjacent information benefits from being tracked separately: the non-sensitive source links cited across multiple workpapers, open review questions that span more than one section, and handoff notes when one staff member passes work to another. These do not belong in the engagement's formal workpaper system, but they are easy to lose in email.

Keeping those references and questions in a project record outside the workpaper system means reviewers and preparers start each session with the same picture. fenbs tracks these without storing any client financial data: the work record holds the references and open questions, not the numbers behind them.

Handing off between preparers and reviewers

Workpaper preparation often spans multiple people. A staff accountant completes a section, a senior reviews it, a manager approves it. When a section is passed up the review chain with unresolved questions, those questions need to travel with it clearly. A handoff note that says "pending client confirmation of the lease term" is more useful than an open comment buried in the workpaper itself.

Tracking those open questions and handoff notes in a separate record, with clear ownership and current status, prevents review items from being accidentally closed without resolution.

How to do this with fenbs

1. Create a free account and set up a project for an engagement or a workpaper preparation area you handle regularly.
2. Add the source references that appear across your workpapers as Links and listings: the relevant ASC topics, the prior-year workpaper references, and the client-provided documents cited.
3. Track open review questions as tasks with clear ownership: which question is waiting on the client, which is waiting on the reviewer, and what the next step is.
4. Use Where we left off to hand off between preparer and reviewer: what is complete, what is outstanding, and what the reviewer needs to know before picking it up.
5. Connect your AI assistant over MCP from Integrations so it reads the open questions and references before starting a session.

For related topics, see AI agents for accounting, AI in accounting and auditing, and AI for forensic accounting.