watsonx Platform for Internal Audit Report Generation & Validation
Co-Author Vivek Salve
Generative AI has the potential to substantially cut down on internal audit expenses by generating standardized reports that adhere to established methodologies, all while enhancing risk management and boosting overall assurance for the organization.
The Internal Audit (IA) function serves as the final safeguard and an essential element of an organization’s governance framework. Its primary objective is to offer independent and objective assurance and services aimed at enhancing value and improving operations. A typical audit report includes both quantitative and qualitative information. However, the qualitative components, such as recommendations, conclusions, trends, and the executive summary, are often problematic due to the subjective nature of the auditor’s judgement, training, and adherence to the internal audit methodology.
Reports that lack clarity and conciseness can be challenging to comprehend, especially when they involve technical details. The excessive use of technical jargon can make the report difficult for non-specialists to understand, leading to additional inquiries from internal and external stakeholders. Moreover, failing to follow the methodology can result in a lack of context, making it challenging to comprehend the implications and significance of the audit results. On the other hand, excessively lengthy or verbose reports can discourage readers from conducting a thorough review, leading to quality issues.
watsonx Solution
Internal auditors can harness the power of generative artificial intelligence (AI) through watsonx to facilitate the creation of audit reports in accordance with their organization’s internal audit (IA) methodology. Generative AI and automation can be utilized to gather and verify data from a Governance, Risk, and Compliance (GRC) platform, and subsequently generate various types of internal audit reports, such as risk-based, ad-hoc, or partial reports.
The planning, scope, background, trends, and recommendations sections of the report can be standardized to adhere to the organization’s IA methodology. For instance, generative AI can succinctly summarize an issue into two to three sentences, eliminate jargon, provide context, enhance grammar, and make the report accessible to non-experts.
Additionally, generative AI can be directed to describe the audit engagement’s coverage, including the jurisdiction, line of business, and location, in a few sentences. Furthermore, generative AI can create a comprehensive executive summary of two paragraphs, complete with sections that summarize recommendations, management response, findings, and an overall conclusion that aligns with the methodology.
Generative AI can also validate the report against the methodology. In this scenario, an initial draft of the report is generated by the AI, followed by revisions or edits made by the auditor. The auditor then engages in a review and challenge process with the business, which may lead to further edits to the report. The AI process will then score the report and present a graphical depiction section by section, benchmarked against the methodology, to identify and rectify any discrepancies before publication.
A Comprehensive Depiction of the Internal Audit Report Development Process: An Examination of the High-Level Flow and Key Components of the Solution
watsonx Internal Audit Platform Accelerates Report Generation and Validation
The watsonx platform is a cutting-edge artificial intelligence (AI) solution that offers a range of advanced features tailored to the needs of internal audit organizations. It is built on IBM’s leading AI technology and can be deployed on any cloud or on-premises.
The watsonx.ai component of the platform allows users to manage the entire lifecycle of generative AI solutions, including tuning, validation, and deployment. It offers access to foundation models from IBM and other sources, enabling the use of expansive language models for a variety of natural and programming language use cases, including report generation. The platform also includes the Prompt Lab tool, which simplifies the prompt engineering process and allows users to quickly start internal audit projects with confidence.
The watsonx.governance component helps users implement responsible, transparent, and explainable AI workflows for both generative AI and machine learning models. It combines the capabilities of Watson OpenScale, AI Factsheets, and OpenPages Model Risk Governance into a single service, and extends its governance provisions to include generative AI assets. This allows users to assess foundation model prompts and machine learning models, construct AI use cases, and monitor lifecycle activities with precision.
Finally, watsonx.data facilitates scalable analytics and internal audit AI initiatives by providing centralized access to data from diverse sources, eliminating the need for migration or cataloging. It includes integrated vectorized embedding capabilities for data preparation and a gen AI-powered conversational interface for simplified data discovery, augmentation, and visualization. The platform also integrates seamlessly with existing databases, tools, and modern data stacks for interoperability.
Conclusion
When it comes to generating and validating internal audit reports, using the watsonx platform can truly revolutionize the way we approach reporting. Plus, it helps ensure that AI initiatives are transparent and responsible. By taking advantage of all the features that watsonx has to offer, organizations can easily navigate the complex world of internal audit. This platform allows users to confidently assess, monitor, and optimize AI workflows, from model tuning to data management. By doing so, watsonx helps organizations comply with regulatory standards and fosters innovation and trust in their AI-driven solutions.
