A practical guide / Auditors
AI in audit: how audit firms can use it
AI can help sort audit queries, prepare clear findings and turn review notes into assigned actions. A good first task is a query tracker for one audit area, built from working papers your team has already prepared.
The FRC’s March 2026 guidance gives examples of board-minute summaries and contract review. This guide proposes a file task: make each open query easy to trace, answer and review. Use the source guidance in the final section to set your firm’s controls.
Use the query promptLink each audit query to its evidence and response.
Give each query a source link
Choose a defined area, such as fixed asset additions. Assemble the test schedule, file links and the team’s existing queries. Ask the assistant to group related points under a stable query number while keeping the source working paper ID.
More on this step
Use these tracker fields: query ID, audit area, source link, finding, details requested, owner, due date and review status. Put a separate link against the response when it arrives. The FRC’s 2025 AI guidance addresses records kept by the firm and on the audit file. Use it when deciding what to retain about the tool and its role.
Read the source: FRC: AI in Audit, documentation guidance
Write findings with a clear basis
Use a short drafting structure: what was observed, which evidence supports it, why the team raised the point and what response is needed. Supply the relevant criterion from the audit team’s work. Ask for wording that separates the finding from the team’s view.
More on this step
For example, replace “asset support is poor” with a request naming the sample items and missing records. Keep the client request short, with working paper detail held in the tracker. Ask the senior which queries belong together before sending the list. That gives the client a focused set of questions to answer.
Turn review comments into actions
Prepare a review-note list with note number, working paper, reviewer’s question, preparer’s response and clearance link. Ask AI to suggest a short action label for each note. Keep the reviewer’s own words available beside the label.
More on this step
ISA (UK) 230 is the FRC standard on audit documentation. Use the firm’s documentation policy when retaining review evidence. As a practical handover, ask the preparer to link the revised work and state what changed. Let the reviewer record whether the response resolves the note. Carry the same note number through that exchange.
Read the source: FRC: ISA (UK) 230, Audit Documentation
Worked example
An additions query that the client can answer
Example with made-up details: the test schedule lists additions A12 and A17. The given pack contains invoice A12 but has no invoice for A17. The team has already recorded the missing support in working paper FA03.
The draft query reads: “Q07, fixed assets, FA03: please send the invoice for item A17 in the test sheet, or explain which document supports the entry.” Assign the request to the client finance contact. Link the reply to its own file. The senior can then assess the evidence without searching a general email thread about assets.
Illustrative example.
Working papers
Keep sample and document references.
Query tracker
State the observation and request.
Review record
Link the response and clearance.
Try a practical task
Prepare a focused audit query list
Use this with a sample test schedule and review notes, following your firm’s tool policy.
Try this in your AI app
ChatGPT
Claude
Gemini
CodexUsing Gemini? The button copies the prompt. Paste it into the new chat.
You are helping prepare an audit query tracker for fixed asset additions. Use the test sheet and the audit team’s notes. Extract each explicit request for further evidence. Retain the sample number, working paper ID and file name.
Draft a client-facing query in neutral language. Distinguish a missing document from a discrepancy the team has already identified. Group possible duplicates for the senior to review, retaining both earlier links.
Produce a tracker with query ID, finding, source link, requested response, owner and status. After it, list draft review-note actions using the reviewer’s note number. Keep each query open for the audit team’s evidence review and decision.
Retain the basis for each revision
When a reply changes a draft finding, link the source and record why it changed. Keep the client reply, any new document and the team’s conclusion in separate fields.
For the first exercise, choose a closed area and compare the draft tracker with the existing queries. Look for missing links, questions that should stay separate and changes to the team’s meaning. Ask a senior to record the points to fix. Use that feedback to refine the draft instructions before a wider trial.
Apply the firm’s audit quality controls
The FRC’s March 2026 guidance says human auditors remain accountable. The controls needed depend on the tool and its use. It also connects AI use with the firm’s quality management obligations. Read it alongside the 2025 documentation guidance cited above.
Use approved tools and access controls for client data. Leave evidence evaluation and the audit opinion with the audit team. Check source links and wording before issuing a query. This plan covers file tasks. Assess any wider use under the firm’s policies, including how it affects professional scepticism and materiality.
Read the source: FRC: generative and agentic AI guidance, March 2026
Questions about this task
Should we start with an entire audit file?
Pick one closed audit area with clear working paper IDs. Use it to refine the tracker fields and query wording before broadening the exercise.
How can we keep review notes useful?
Retain the original question, link the preparer’s revised work and record the reviewer’s clearance against the same note number.