--- ## Slide 0 — Title 「Hello everyone. My name is Goo Kim, and I'm a Senior Researcher at the Board of Audit and Inspection of Korea. Thank you for the invitation to the 2026 IsDB Group Annual Meetings. Today, I'll be talking about our 'Digital Audit Initiatives' — what we built, and what we learned.」 --- ## Slide 1 — Contents 「Here is what I'll cover today. First, why digital transformation. Then, BAI's digital strategy — our two tools, BARON and AuditTalk. And finally, our key takeaways. Let's start.」 --- ## Slide 2 — Why Digital Transformation? 「So, why did we need to change? The growing complexity of the digital age calls for a new audit paradigm. Public data has exploded. The exponential growth of digital government records and transactions has made manual review impractical. Cross-agency data flows and interconnected digital systems have made policy implementation increasingly complex and difficult to trace manually. And auditor capacity remains limited, while the volume and complexity of audit targets continue to grow. Digital Government, Data Explosion, Audit Complexity, Need for Innovation. This is why we started our digital transformation.」 --- ## Slide 3 — BAI's Digital Strategy 「So, here is BAI's response. We built two complementary systems covering the entire audit cycle. BARON — for audit data collection, analysis, and data-driven audit insights. And AuditTalk — an A.I. agent for audit case search and support for risk analysis. Let me explain each one.」 --- ## Slide 4 — Part 1: BARON 「Part 1. BARON. Best Audit & Inspection System for Rule-based Observation Network.」 --- ## Slide 5 — What is BARON? 「What is BARON? BARON is a system that collects and analyzes audit data — to support audit planning, risk identification, and audit execution. Why did we build it? As government operations become increasingly dependent on IT systems, auditors must understand and effectively leverage those systems. Stricter data regulations require auditors to establish secure mechanisms for accessing audit data, rather than relying on ad hoc on-site requests. And remote auditing requires digital capabilities to collect, analyze, and utilize audit data. BARON covers four steps: Data Collection — audit-focused data from financial, administrative, and spatial government systems. Data Analysis — advanced analytics tools, including statistical modeling, anomaly detection, and scenario analysis. Visualization — charts, graphs, and interactive maps for intuitive data visualization. And Advisory Services — database analysis, audit scenario development, and specialized support for audit teams.」 --- ## Slide 6 — BARON Key Services 「BARON provides four key services. Financial Data Analysis — budget, expenditure, subsidies, and public assets. Administrative Data Analysis — building permits, vehicle records, and citizen records. Spatial Data Analysis — map overlay, and buffer and radius analysis. And Info Systems — comprehensive database schema information for audited entities.」 --- ## Slide 7 — BARON Analysis Tools 「Auditors can directly access database schemas and conduct in-depth analysis using built-in tools. Information System DB provides the full DB schema for ALL audited agencies with sample data for BARON-linked systems. S.Q.L. Modeler is a visual drag-and-drop interface for building complex multi-table queries without coding. And BARON Sheet is an Excel-like self-service analytics tool that enables auditors to analyze BARON data directly.」 --- ## Slide 8 — BARON Financial Data 「In terms of scale — BARON is linked to 35 agencies, 22 sectors, and 53 systems. 643 data categories are available for auditing. This includes national finance, local finance, education, public institutions, procurement, and research funds and so on.」 --- ## Slide 9 — BARON Agency View 「BARON also provides an agency-specific dashboard — integrating key indicators, operational status, legislation, financial data, administrative information, and audit knowledge in one place. For example, for the Ministry of Land, Infrastructure and Transport — you can see all of this at a glance.」 --- ## Slide 10 — BARON: How It Works 「Here is how BARON works in practice. Step 1 — the audit team identifies audit focus areas, defines data requirements, and submits data extraction requests. Step 2 — the Data Analysis Center provides the requested data and conducts data-driven analyses. Step 3 — the audit team uses the data to conduct the audit, and provides feedback. Step 4 — the data is continuously accumulated and reused for future audits. We also provide core audit data, consulting, and tailored analytical support for audit teams.」 --- ## Slide 11 — BARON in Action 「Due to time, I'll skip this part — please take a look at the slides.」 --- ## Slide 12 — Part 2: AuditTalk 「Now, Part 2. AuditTalk. Audit A.I. Agent. Case Search, Risk Analysis, Evidence Discovery.」 --- ## Slide 13 — What is AuditTalk? 「What is AuditTalk? AuditTalk is an A.I.-powered audit agent with two core exploration modes. Auto Search — enter a keyword, and AuditTalk identifies high-risk audit areas and recommends search queries for related issues, research, and audit cases. For example, type "Defense", select a related field, receive recommended queries. Click a link to view results instantly. Manual Search — type slash and select a search mode. For example: slash news, "Tell me about recent policy issues in the defense industry." Or: slash korea, "Tell me about A.I.-related audit cases." AuditTalk supports six search modes: slash korea for Korean audit cases, slash usa for US G.A.O., slash uk for UK N.A.O., slash papers for academic research, slash research for policy reports, and slash news for real-time news.」 --- ## Slide 14 — AuditTalk Architecture 「AuditTalk is built on RAG — Retrieval-Augmented Generation — powered by six curated knowledge sources. It pulls from BAI audit reports, US G.A.O., UK N.A.O., academic papers, research reports, and real-time news. The output includes A.I.-generated analysis, source citations, and case deep-dive Q&A. Citations ensure the A.I. output is traceable and reliable.」 --- ## Slide 15 — AuditTalk UI 「AuditTalk is accessible via web browser — no special setup needed. The main interface lets you select a search mode. In Auto Search, a keyword gives you A.I.-recommended risk areas and queries. You can click any query for instant results, and click any case for a detailed Q&A.」 --- ## Slide 16 — AuditTalk Use Case 「Here is a real example. Type slash korea, "A.I. Audit Cases." AuditTalk returns a summary and a list of 8 detailed cases instantly. Click any case — deep-dive Q&A. Relevant laws, recommendations, and findings are available on demand. Research time reduced from weeks to minutes.」 --- ## Slide 17 — Lessons Learned 「So, what did we learn? From BARON — Data-driven auditing is now feasible at scale. Cross-database analysis reveals hidden patterns. It reduces reliance on manual sampling and accelerates evidence collection significantly. From AuditTalk — A.I. dramatically shortens pre-audit research. International precedents improve audit quality. Natural language lowers the technical barrier. And citations ensure the A.I. output can be verified and trusted. Common benefits — Faster analysis with better evidence. Enhanced risk identification. More strategic use of auditor expertise. And stronger public accountability outcomes.」 --- ## Slide 18 — Conclusion 「To wrap up. BARON brings data analytics. AuditTalk brings A.I. intelligence. Together — smarter public auditing. BAI Korea is committed to advancing audit quality through digital innovation — ensuring accountability, transparency, and public trust in the digital age.」 --- ## Slide 19 — Thank You 「Thank you very much. If you have any questions, please feel free to reach out by email — my contact is on the screen.」 ---