Audit-Ready Faster: AI for PBC Lists & Auditor Requests

By Christian Matumona, Kyootek AI Financial Controller

A PBC — prepared-by-client — list arrives at the start of every audit cycle with forty to eighty discrete requests, ranging from "trial balance as of period end" to "supporting documentation for the largest five customer contracts." Some items you already have ready. Some require pulling together documentation that exists but isn't organized. A few are genuine gaps that need to be created from scratch. The first hour with a PBC list is usually spent figuring out which category each item falls into — and that triage step is where AI removes real time.

This is the deep dive on audit preparation referenced in the Complete Guide to AI for Financial Controllers.

The triage problem

Most of the friction in early audit response isn't producing documentation — it's figuring out, item by item across a long list, what you already have, where it lives, and what's actually missing. Controllers who've been through several audit cycles develop an intuition for this, but intuition doesn't scale to a new team member or a particularly long PBC list, and it still takes real time to apply even when you have it.

AI is useful here specifically because matching a list of requests against a list of available documentation is a pattern-matching task — exactly what these tools do well, provided you give it both lists in a structured form.

The triage prompt

You are helping triage a PBC (prepared-by-client) list against existing documentation.

PBC list (item number and request description):
[PASTE THE FULL PBC LIST]

Documentation we currently maintain (description and location/owner):
[PASTE A LIST OF YOUR STANDARD DOCUMENTATION — e.g. "Trial balance — Finance shared drive, updated monthly", "Customer contracts >$100k — Legal repository", "Fixed asset register — maintained by Controller"]

For each PBC item, classify as:
- Ready — documentation exists and matches the request as described
- Needs assembly — documentation exists but needs to be compiled, extracted, or reformatted to match the request
- Gap — no existing documentation addresses this request; needs to be created

Sort the output by category, with Gap items first. For Gap items, briefly note what would need to be created.

Run this as soon as the PBC list arrives. The output reorders a forty-item list into something genuinely actionable: a short list of real gaps to start on immediately, a list of assembly tasks to delegate or schedule, and a (usually larger) list of items that are simply ready to package and send.

Why "Gap" items deserve attention first

The instinct under audit-cycle pressure is to clear the easy items first — it feels productive. The better sequencing is the opposite: genuine gaps take the longest to close and are the ones most likely to cause a delay if discovered late. Surfacing them on day one of the cycle, rather than as they're identified piecemeal through auditor follow-up questions, is the single biggest timeline benefit of doing this triage upfront.

Drafting documentation narratives for assembled items

For "Needs assembly" items, AI can also help draft the cover narrative that typically accompanies a PBC response — a short explanation of what's included and how it was compiled — once you've gathered the underlying documents.

Draft a one-paragraph cover note for a PBC response.

Request: [PASTE THE SPECIFIC PBC ITEM]
What's included: [LIST THE DOCUMENTS YOU'RE PROVIDING]
Compilation method: [BRIEF NOTE ON HOW THE DOCUMENTATION WAS ASSEMBLED, e.g. "extracted from the GL by account range, reconciled to the trial balance"]

Write a clear, factual cover note an auditor can read before reviewing the attached documents. State only what's listed above — do not characterize whether the documentation is sufficient or complete.

That last constraint matters. Whether a response is complete and sufficient is an audit judgment, not something AI should imply on your behalf.

What stays entirely with you

Every item that goes to an auditor is selected and reviewed by a qualified professional before it's sent — AI's role ends at triage and drafting. Decisions about what to disclose, how to characterize a judgmental item, or how to respond to a follow-up question that touches a sensitive area remain audit and, where relevant, legal decisions. The same data handling discipline that applies elsewhere in the close applies here too: see the guardrails in the Month-End Close Playbook for what should never be pasted into an AI tool, particularly during a period when sensitive items are actively under review.

The compounding effect across an audit cycle

A well-triaged PBC list at the start of an audit cycle changes the shape of the whole engagement — fewer surprises in week three, less scrambling on follow-up requests, and a clearer picture of where the real risk sits. That's a direct, practical expression of the governance-minded use of AI described in What Is an AI-Augmented Financial Controller?: AI accelerates the organizing work so you can spend the audit cycle on the parts that actually require your judgment.