Notion-based audit documentation system paired with a reusable library of AI prompts for engagement work.
AVIONICS STACK
FLIGHT MANUAL — DEEP DIVE
This project explored the use of large language models to accelerate the drafting of audit documentation — working papers, management letter points, and audit committee reports. The objective was not to automate the auditor's judgement but to reduce the friction of translating judgement into written form.
The approach uses carefully designed prompts that supply the model with the audit context: the engagement facts, the applicable standards, the specific procedure performed, and the evidence obtained. The model generates a first draft, which the auditor then reviews, revises, and signs off. The auditor remains the author; the model is a drafting assistant.
The quality of the output depends almost entirely on the quality of the prompt. A vague prompt produces a vague draft; a precise prompt that specifies the audience, the purpose, and the required content produces a draft that is close to final. The skill of prompt engineering is the skill of specifying what you want.
The time savings are real but secondary. The primary benefit is that the auditor's time is spent on review and refinement rather than on facing a blank page — a shift that improves both efficiency and the quality of the final document.