From Two Weeks to Minutes

If document-heavy research consumes your time or repetitive tasks slow you down,
keep reading.

Much of the time spent on compliance reviews is not spent making decisions; it's preparing the information needed to make them. By using AI and automation to perform much of this document-intensive work, Internal Audit has streamlined a process that once required approximately two weeks to about one day, a 93% reduction in time spent. Auditors retain responsibility for review and validation.

THE CHALLENGE

The most time-consuming part of a compliance review often occurs before the review itself begins. Auditors locate the appropriate policy set, read lengthy statutes and policies, compare individual requirements, identify where language aligns or falls short, and assemble findings for review. Internal Audit explored AI not to replace audit judgment, but to reduce searching, sorting and first-pass comparison so staff could focus on higher-value analysis.

THE SOLUTION

The team first experimented with the standard Copilot experience in Microsoft 365. It could respond to requests, but the answers were not consistently accurate or controlled enough for this compliance task. That limitation shaped the next step: Internal Audit used Copilot Studio to create a policy review-focused agent and Power Automate to build a repeatable process around it. Copilot Studio grounded the agent in selected policy and guidance sources. Power Automate connected the activities, including retrieving files, preparing documents, initiating analysis, routing reports for approval, and compiling approved results, into one coordinated workflow.

IMPLEMENTATION CONSIDERATIONS

Some statutes can run hundreds of pages, while the AI processing step has document-length limits. The workflow isolates relevant sections and divides large files when necessary. Designated folders and triggers then move each completed action into the next stage, turning an AI capability into an operational process.

HOW IT WORKS

Once the legislation and policy files are prepared, the AI-assisted workflow evaluates statutory requirements against the existing policy. It generates draft reports that identify potential matches, indicates when a requirement does not appear to be addressed, and captures supporting information such as policy revision dates. The tabular output is structured for auditor review and is not treated as a final conclusion.

THE IMPACT

Draft comparisons can be generated within minutes. The practical gain is a shift in effort: less time locating and assembling information, and more time assessing whether the analysis is complete, accurate, and relevant.

HUMAN OVERSIGHT

Every AI-generated report reaches a human checkpoint. An auditor reviews each report, returning rejected drafts to the AI step and allowing only approved ones to move forward. Once it advances, Power Automate consolidates approved information into an Excel summary for final review. Auditors remain responsible for the audit product and professional judgments drawn from the evidence.

LESSONS LEARNED

Internal Audit developed the workflow in-house, with team members learning the tools through research, experimentation, and hands-on experience. Aiden Ho, Auditor II, explained that he did not begin with a coding or automation background. He learned through research, experimentation, and guidance built into Microsoft’s tools. That matters because the people closest to a process often understand where the repetitive effort and decision points actually sit. With time to learn and test, subject-matter knowledge can become the foundation for a practical AI solution.

THE TAKEAWAY

Internal Audit’s experience shows that meaningful AI adoption begins with a clear process, trusted information, human oversight, and defined outcome. By redesigning the repetitive parts of compliance review, the team reduced a roughly two-week effort to draft analysis in minutes and an auditor-reviewed process of about one day, while keeping professional judgment exactly where it belongs: with the auditors.

For the UH community, the broader lesson is clear: AI can speed up repetitive review work where significant time is spent reviewing documents and preparing information for analysis while people remain accountable for the final decisions.

AI Solutions