How can accounting firms manage audit request lists and follow-ups with AI?
Give a ChatGPT Skill your approved audit request list template, the engagement details and the items the engagement team has selected from the audit programme. It drafts a client request list in plain language, with each item’s period, preferred format, client owner and due date. As documents arrive, staff update the tracker and the Skill drafts follow-ups listing only what is still outstanding or incomplete. The audit senior checks every list and message before it is sent. Whether evidence is sufficient remains an audit judgement made by your team.
Why audit requests get lost
Audit fieldwork depends on what the client provides, sometimes called a prepared-by-client or PBC list. In many firms that list is a long spreadsheet copied from last year, sent as an attachment and then chased through email threads. Clients do not know which items are urgent, who on their side should answer, or what “GL detail” means. The audit team loses days working out what has arrived. A request list Skill writes each item so the client’s finance team understands it, and keeps follow-ups focused on what is actually missing.
What each request item contains
Swipe to read every column, or focus the table and use the arrow keys.
| Field | Drafted from | Who decides |
|---|---|---|
| Item, in plain words | The procedure in your audit programme the engagement team selected | Audit senior confirms the wording asks for the right thing |
| Period and cut-off | The engagement’s financial year-end | Audit senior |
| Preferred format | Your template, such as an Excel export or signed PDF | Audit senior |
| Client owner | The client contacts list the team keeps | Client confirms at the planning meeting |
| Due date | The fieldwork plan | Audit manager |
| Status | The tracker staff update as documents arrive | Audit team reviews each item received |
A worked example
Tracker extract (synthetic)
Client: sample trading company (mainland LLC) · year ended 31 Dec
Request list v2, sent 12 Jan · fieldwork starts 9 Feb
1. Trial balance at 31 Dec ............ received
2. Bank statements, Dec, all accounts .. received (2 of 3)
3. Bank confirmation authority letter .. outstanding
4. Fixed asset register with additions .. received, additions
not listed
5. Year-end inventory count sheets ..... outstanding
6. Top 10 customer contracts ........... outstanding
Follow-up drafted by AI for the audit senior
Dear finance team, Thank you for the documents so far. To keep fieldwork on 9 February, we still need the following by 30 January: 1. December statement for your third bank account (the one not yet received) 2. The signed bank confirmation authority letter (template attached again) 3. A list of fixed asset additions during the year, with invoices 4. The signed inventory count sheets from 31 December 5. Your 10 largest customer contracts Items 2 and 4 usually take longest, so it helps to start with those. For the audit senior: item 2 bank account not named in the tracker. Confirm which account before sending.
Synthetic client, items and dates for illustration only.
How it fits your audit process
- The audit programme stays the source. The engagement team selects which procedures apply. The Skill turns them into client requests; it does not decide scope, materiality or what to test.
- Your templates, with versions. The request list template and follow-up wording are approved by the audit manager. When they change, the new version is approved and the Skill is refreshed; the old version is withdrawn.
- Same approach as tax document chasing. Firms that already chase VAT and corporate tax documents can reuse the pattern. See chasing client documents for VAT and corporate tax.
Guardrails
- The audit senior checks every list and follow-up before it is sent. AI drafts can leave out an item, merge two requests, get a date wrong or add an item that is not in the programme. Instructions reduce this risk but do not guarantee a complete or correct list.
- Receiving a document is not the same as accepting it as evidence. The audit team reviews each document; the Skill marks items as received only when staff record them, and never concludes on sufficiency or quality.
- Audit judgements, independence and the opinion stay with the engagement partner and your firm’s methodology and quality procedures.
- Client information is confidential. Start with sample or anonymised lists; real client data belongs in your approved ChatGPT workspace with the access, retention and other duties your firm applies.
- Manual input is the default. Staff update the tracker and paste it in. Reading from your audit software, client portal or email directly needs a separately agreed, configured and tested integration.
An audit request list Skill can start as a single Skill for AED 97, built on your own template. Audit request lists are part of RUN at AED 15,000, for larger firms or several service lines that also want the partner’s weekly brief and approval rules. See the pricing page and what AI costs for an accounting firm.
Common questions
Can AI create an audit request list for a client?
Yes, as a draft. From your approved template and the procedures the engagement team selected, it drafts each request in plain language with period, format, owner and due date. The audit senior checks it before sending.
Can it follow up on outstanding audit documents?
Yes. From the tracker your staff keep, it drafts follow-ups listing only what is still outstanding or incomplete. Staff send them.
Does AI decide whether the audit evidence is enough?
No. It only tracks what staff have recorded as received. Reviewing documents and deciding whether evidence is sufficient stays with the audit team and engagement partner.
Does it connect to our audit software?
Not by default. Staff paste the tracker. Reading from audit software, a client portal or email needs a separately agreed, configured and tested integration.