Name in original language
AuditPesquisa – Sistema de Inteligência Artificial para Apoio à Auditoria Interna Governamental
Initiative overview
Before AuditPesquisa, searching for information in Brazil's internal government audit reports was manual, time-consuming, and inefficient. The CGU's Federal Office of Internal Control (SFC) maintained over 3,000 audit reports but lacked a system to interpret textual content, identify patterns, or connect similar findings across audits. As government demands for faster, evidence-based analysis increased, there was a critical need for an intelligent solution that could understand context and meaning—going far beyond simple keyword searches.
The AuditPesquisa initiative pursues five primary objectives: 1. Facilitate Access to Audit Knowledge: Enable auditors and government managers to rapidly access insights embedded in 3,000+ historical audit reports through semantic understanding rather than keyword matching. 2. Reduce Research Time: Significantly decrease the time required to search for audit information and prepare new audit procedures and recommendations. 3. Promote Knowledge Reuse: Encourage the reapplication of prior audit analyses, findings, and methodologies in future auditing work to build on institutional learning. 4. Create an Intelligent Repository: Consolidate audit reports in a searchable, semantically indexed database with advanced filters (by agency, audit type, time period, findings, and recommendations). 5. Foster AI Innovation in Public Administration: Demonstrate the value and feasibility of generative AI for core government functions, building a culture of responsible AI adoption within Brazil's internal audit community.
Results, outcomes and impacts
Since its launch in October 2025, AuditPesquisa has registered hundreds of interactions by almost 200 users across CGU, at a total operational cost of just USD 36.23 — demonstrating high impact at minimal expense. More than 3,000 public audit reports were indexed and semantically vectorised, significantly reducing information retrieval time for auditors. Usage metrics and user feedback serve as the primary measurement tools. Expected future impacts include broader adoption across federal oversight bodies and replication to other document-intensive government functions.
