Related posts
See all
Key insights
- Review notes are how supervision gets recorded in the audit file. When an agent drafts first, the notes arrive earlier and ask about evidence, not tie-outs.
- The amended Auditing Standard (AS) 1215 gives firms 14 days after report release to finish the file, starting December 2026. Review can't slide to the end anymore.
- Proofreading a workpaper instead of evaluating it is already a known failure mode. Agent-drafted work looks polished faster, so source citations, exception flags, and a traceable record have to carry more weight.
Every engagement ends the same way: a stack of review notes, a partner asking whether they've all cleared, and a report deadline that doesn't care either way. The notes are where supervision actually gets recorded, and where the coaching, the caught misstatements, and the last-minute rework all live. Change who writes the first draft of the workpaper, and every one of those notes changes with it.
When an agent produces the draft instead of a first-year staff member, the notes arrive earlier in fieldwork, land in front of a different reviewer, and stop asking about tie-outs. That shift matters more now that AS 1215 gives firms 14 days after report release to close the file. This article covers what review notes are supposed to do, why the traditional timing is running out of room, and how the first-pass function changes when an agent produces the draft.
What review notes are supposed to do
Review notes are the written back-and-forth between preparers and reviewers on a workpaper: a question, a correction, a request for more support, and the resolution recorded against it. They make supervision visible in the file.
The engagement partner owns the conclusion that the audit was performed as planned, that significant judgments hold up, and that the evidence supports the opinion. Teams reach that conclusion through layers of review, and review notes carry questions down the team and resolution back up.
Notes have always carried a second job alongside quality control: they're how junior staff learn. A note that says "your tie-out missed the subsequent receipt, here's why that matters for the assertion" teaches something a checklist never will. But reviewer grading often focuses more on what reviewers catch than on what they teach, so the coaching half has always been inconsistent. It depends on who's reviewing and how much time they have.
What happens to the notes after they clear is its own question. PCAOB's documentation guidance says reviewers may annotate workpapers with clarifying questions and edits during review, and resolved review comments and open-items lists need not remain once the file is complete. The long-standing practice is to confirm the notes cleared, verify the requested work made it into the workpapers, and then discard the notes themselves. Retained notes that were never properly resolved can create avoidable litigation risk. The work stays in the file; the conversation about the work can be discarded after proper resolution.
How review timing traditionally gets planned
The plan and the file don't usually agree on when review actually happens, and the gap between them is about to get more expensive.
The plan says review during fieldwork; the file says otherwise
On paper, review is a two-step structure: a detailed pass over staff work, then a higher-level partner pass focused on significant risks, judgment areas, and any contentious matters. In practice, both passes drift to the end of the engagement, because that's when the workpapers finally exist to review.
The reason is staffing math. When senior engagement team time is low relative to staff time, partners and managers don't have the hours to supervise review work or evaluate audit judgments as fieldwork unfolds — so review waits until there's something finished to look at. That's the imbalance the firm and engagement metrics rulemaking is trying to make visible. It's also what produces the late-stage EQR surprise, where significant issues only come to light when the partner hands over a finished product days before the deadline.
The 14-day window removes the cushion
The old workflow survived because there was slack after report release. That slack is shrinking. The amended AS 1215 cuts the documentation completion period to no more than 14 days after the report release date, effective December 15, 2026. Supervisory reviews still must be complete before the report goes out, and the shortened window is aimed squarely at inspection readiness.
Review notes written near the end of an engagement were always a quality risk. Soon they'll be a compliance risk too. High and increasing deficiency rates tied to engagement quality reviews show the review chain is already straining under the current timeline.
Where the first pass traditionally falls on humans
The first detailed pass has always belonged to seniors and managers. Staff prepare the workpaper; the senior ties the support as part of a line-by-line review and writes the notes; the staff member clears them; the senior verifies the clearing; and points needing partner attention accumulate upward through an issues memo. That creates multiple trips through the same workpaper before anyone with signing authority sees it.
The pattern does two jobs at once. It consumes senior and manager hours with clerical verification, exactly the people the PCAOB worries have too little time for judgment. It also trains the profession's next generation of reviewers. Reading messy staff work and articulating what's wrong with it is how a senior becomes a manager.
PCAOB's 2024 inspection update put the aggregate Part I.A deficiency rate at 39%, improved from 46% in 2023. Inspectors still found significant deficiencies in roughly four of 10 inspected audits. That's what an all-human review chain produces when it runs at capacity, and the 14-day window makes that capacity smaller.
What changes when an agent does the first pass
When the first draft comes from an agent instead of a first-year staff member, the review note becomes a more important artifact, not a less important one. It's what records the questions raised on the draft and the resulting correction or escalation.
The note's timing and audience both move
Agent-drafted work can be reviewed as evidence lands, rather than waiting for an end-of-fieldwork batch. On Fieldguide, Field Auditor executes test procedures as evidence comes in and documents results with citations and exception flags. Agent Review Experience gives preparers a dedicated workspace to review and append to agent output before it passes to the manager.
That restructures the note cycle in two ways:
- The detailed first-pass review can happen earlier in fieldwork instead of near the deadline, which is often the most practical way to work inside the shortened documentation window.
- Notes on agent output split into two kinds: judgment questions that still travel up the human chain, and correctable feedback that goes back into the workflow as a rerun rather than a note a staff member clears later.
Agent runs also produce a Trace showing inputs, outputs, and reasoning, so reviewers can see how the draft was built. UHY reported 20–30% engagement time reduction with this kind of workflow, with some tasks going from three hours to 15 minutes.
The proofreading trap gets more tempting
Reviewers may proofread agent-drafted work instead of evaluating it. The pattern isn't new. PCAOB inspectors have already flagged engagement quality reviews that amounted to little more than proofreading the final draft rather than substantive evaluation.
That happened with human-prepared work that looked finished. Agent-drafted work looks more finished, faster. Avoiding rubber-stamping AI requires more than content expertise; reviewers also need to understand the system's limitations and failure modes.
Engagement team members using AI tools remain responsible for the results, and supervisors are expected to apply the same diligence reviewing AI-assisted work as any other work. That's why agent output inside an engagement file needs to carry its evidence with it: direct source references on every conclusion, exception flags on anything that doesn't reconcile, and a Trace of the inputs and reasoning the reviewer can inspect. Field Auditor is built to produce that record.
Where review time moves
The hours that remain shift from tie-out verification to the two questions the audit evidence standard already puts in front of the reviewer: whether the evidence is sufficient and appropriate, and whether its nature, source, and circumstances make it reliable. That applies whether a human or an agent assembled the support. From there, the most inspection-sensitive questions tend to sit in the exceptions and higher-risk areas the agent flags, and that's where review hours concentrate.
Staff development moves too
Review notes used to answer the staff-development question quietly, through the first pass on staff work. When AI-generated work looks complete, junior employees may struggle to judge whether it is any good — which is exactly why the agent-output review itself becomes the training ground.
A staff member who tests an agent's workpaper against evidence, challenges its exception calls, and explains the resolution to a manager is doing higher-order work earlier in their career than the tie-out treadmill ever offered. That requires firms to treat agent-output review as a designed skill instead of something staff absorb by accident.
Firms that treat the review note that way will be better positioned to manage the compressed deadline. Firms that keep end-loading review will find the new deadline less forgiving.
The platform built for agent-led execution and human review
Fieldguide is an end-to-end AI-native platform, purpose-built for audit and advisory, with the Agent Workforce, methodology depth, and audit-grade rigor firms need to run engagements this way. Agent output, reviewer feedback, reruns, and sign-offs all live in one place instead of scattered across email and tracking spreadsheets, so review discipline holds inside the 14-day window. If your review notes are still piling up near the end of every engagement, book a demo and take a look at what the first pass could look like instead.
Amanda Waldmann
Increasing trust with AI for audit and advisory firms.
