Audit File Review Agent · Use case
How can an audit firm use an AI agent to review an audit file before the partner's review on its own infrastructure?
An audit file review agent reads an audit file against the firm's methodology and lists working papers with a missing sign-off, an open review note, a reference that leads nowhere or a conclusion with no evidence behind it, each with its reference. The engagement manager and the partner review and conclude. The agent signs nothing, concludes on nothing and changes no working paper. It runs on the firm's own infrastructure.
A file holds every working paper.
Each one is read by hand.
- ✗ 01An audit file holds working papers, review notes and cross-references across areas and team members, and open items are found by reading.
- ✗ 02A missing sign-off or an open review note that is found late is a problem for the engagement and for the firm.
Where it stops for a person
Always. The engagement manager and partner review and conclude.Always. Whether evidence is sufficient and appropriate, what the conclusion is and whether the file is complete are the judgement of the engagement manager and the partner. A working paper that is not on the agent's list has not been found sound.
It lists what is open.
The partner concludes.
Step · 01
The file is connected through your own APIs, or the working papers are uploaded.
Step · 02
The agent reads the firm's methodology for that type of engagement.
Step · 03
It reads each working paper, its sign-offs, its review notes and its references.
Step · 04
It lists each working paper with a missing sign-off, an open review note, a broken reference or a conclusion with no evidence, with its reference.
Step · 05
The engagement manager and the partner review the list, review the file and conclude.
Reads from · connectors
File Upload
Your own APIs
The audit file, read through your own APIs or as uploaded files, and the firm's own methodology and review checklists, indexed as a knowledge source with each reader's permissions kept.
Acts through · MCP servers
Nothing. It signs no working paper, clears no review note and writes nothing to the audit file.
What you set up
- If your audit platform gives you an API description, import it and keep only the operations that read. If not, export the working papers as files.
- Index the methodology and review checklists as a knowledge source.
- Build the agent in Katonic Studio and test it on a closed file where you know the open items.
- Publish it to engagement managers before the partner's review.
Where it runs
The agent, its knowledge store and the model run on infrastructure you own or control. You choose the model and serve it yourself, so questions, documents and answers are not sent to a public AI service. Audit files and clients' financial information stay inside the firm.
How you get it today
This page is labelled Use case: the agent is built in Katonic Studio. You describe the job, connect what is named above, add test questions and release the agent once it passes them. When the Catalog holds a tested template for this job, this page will carry the Template label and show how to take it.
Does it sign off working papers?
No. It signs nothing and clears no review note. The engagement manager and the partner review and conclude.
Does it judge whether the evidence is enough?
No. It lists conclusions with no evidence behind them. Whether evidence is sufficient and appropriate is for the auditor.
Does it need a particular audit platform?
No. It reads working papers as uploaded files or through the API your platform gives you. It names no product, and nothing here is a tested integration.