5 min read

Quick-Draw Reporting: From Raw Data to Readable in Record Time

Quick Read

This case study showcases how an AI-powered solution transformed audit reporting from a time-consuming struggle to an efficient process. It follows fictional auditor Allison, who previously battled with scattered information across platforms, tedious evidence gathering, and laborious citation tracking.
The product team built a generative AI tool that automatically analyzes thousands of pages of audit evidence, extracts key findings, and maintains source document links throughout the process. Implemented as a Microsoft Word plugin, the solution reduced report compilation time by 80%, decreased citation errors by 70%, and doubled the speed of identifying issues—demonstrating how thoughtfully designed AI can solve real professional pain points while delivering substantial business value.

Meet Allison

Allison's blank document stares back at her, the cursor blinking like a timer counting down billable hours. "Where should I start gathering all these findings?" she sighs, five minutes into her standoff with an empty draft.
This isn't just Allison's problem. Across audit firms, professionals drown in information scattered across platforms and formats — client details, meeting notes, and evidence documents living in digital diaspora. The struggle continues after completing audit fieldwork, when findings have to be reported back to the client.

Fast and Factually Correct

"The ask is simple — we need a way to trace findings back to their original source," says Allison. Client questions and partner reviews demand instant evidence to back up every claim.
Key details of the issues, supported with summary generated by AI
Key details of the issues, supported with summary generated by AI
When engagements grow complex, team members must become reference librarians on the fly, pulling precise citations from a digital labyrinth of documents. For each finding, Allison needs to recall exactly which file, which page, which paragraph contains the supporting evidence.
This mental mapping becomes unsustainable. What should be a simple "here's where we found this" turns into a frustrating treasure hunt through folders and files — the digital equivalent of searching for a specific sentence in a library without an index.
Breakdown of a issue point raised
Breakdown of a issue point raised

Designing for Tomorrow

To solve Allison's challenges, the product team explored Generative AI as a potential solution. During initial conversations with auditors, we discovered their understanding of emerging technologies rarely extended beyond basic awareness ("I know ChatGPT").
When working with such emerging technologies in digital product design, we found the traditional design thinking narrative needed to shift. Users often struggled to articulate what they wanted from AI because the technology's capabilities weren't well communicated to them. This created a fundamental gap in the typical user research approach – how can users express needs for solutions they don't yet understand?
Reverse double diamond, where we start by exploring
Reverse double diamond, where we start by exploring
The reverse double diamond allows us to deliver first, rather than waiting for complete requirements. We engaged with a small group of dedicated users who worked closely with us during proof-of-concept development. Our process followed this sequence of understanding:
  • Current journey on how they conduct their audit processes
  • Friction and challenges faced within these processes
  • How AI/ML could help elevate the existing processes
  • Ways to trigger the AI-centric actions
By starting with delivery and refining through collaboration, we could introduce capabilities users might not have initially requested but that addressed their underlying needs.
Domain knowledge discovery
Domain knowledge discovery
It takes an equal contribution from both design and technology to achieve a good product experience for AI.
We conducted numerous feedback sessions to understand how they are doing with our early-stage solution and way to improve both the experience and efficiency of the Generative AI model.
NPS from the beta testers
NPS from the beta testers

Acheive The First 40% in Minutes

The solution leverages advanced natural language processing to analyze thousands of pages of audit evidence, automatically extracting key findings while maintaining links to source documents throughout the process. This isn't just a search function—it's a comprehensive knowledge layer that understands audit contexts and relationships.
I used to spend entire afternoons just gathering material for a single section, now I complete full first drafts before lunch, with available issues properly backed and traceable. — Allison
Here's how it works for Allison:
  • When preparing her draft, she inputs current fieldwork findings, and the AI generates a curated list of suggested findings for review, optimally combining them for clarity
  • Each suggestion includes a complete paragraph explaining the finding in detail
  • With a single click within the draft, she can verify sources and incorporate citations directly through the plug-in
  • The system automatically handles report formats and compliance standards, managing structure and formatting so Allison can focus exclusively on delivering valuable insights
The Microsoft Word plug-in features a clean sidebar for requesting and reviewing clarifications without disrupting workflow. This add-in integrates seamlessly with her familiar tools—no switching between applications or learning complex new systems required.
Microsoft Word plugin, with AI assisted for fast clarifications
Microsoft Word plugin, with AI assisted for fast clarifications
The impact has been substantial and measurable after a limited beta testing:
  • Reduced reports compilation time by 80% (from 16 hours to just 3 hours for reviewable drafts)
  • Decreased citation errors by 70% during partner review
  • Doubled speed on identifying the required issues to be raised in the Management Letter.

Organization-wide Adoption

While the solution initially targeted risk auditors, its success has accelerated adoption plans across the organization. Our implementation roadmap now includes:
  • External auditors (Q4 2025)
  • Internal auditors (Q2 2026)
Each new domain requires specialized training data and domain-specific citation models developed collaboratively with subject matter experts, incorporating prior years' documents to train the model for a broader audience.

Notes

The Allison in this story is fictional, but her challenges are real. As an independent contributor AI Experience Designer in the product team, I've designed the solution using generative technology that transforms how auditors compile reports.
Mock data replaces actual information to safeguard confidentiality. The value delivered to audit teams is 100% authentic.