An Agentic Software Framework for Data Governance under DPDP

Adaptive compliance beats hard-coded rules: Traditional data governance tools use static, inflexible rule systems that struggle when laws change or new situations arise. This framework uses AI agents that can dynamically adapt to policy updates, making compliance more responsive to evolving legal contexts like India's DPDP Act.

Transparency in AI decision-making is essential: The paper addresses the "black-box" problem where compliance decisions happen invisibly in monolithic systems. By using specialized agents (KYU Agent for user trustworthiness and Compliance Agent for data sensitivity), the framework makes governance decisions more explainable and traceable—critical for responsible AI.

Multi-domain applicability demonstrates real-world relevance: The framework was tested across ten diverse sectors including healthcare, education, and e-commerce, suggesting that agentic approaches to compliance can work across different industries and data types, not just in isolated use cases.

User understanding and data sensitivity reasoning matter: The framework combines semantic understanding of users (through the KYU Agent) with reasoning about how sensitive different data is (through the Compliance Agent), recognizing that effective data governance requires understanding both who is using data and what kind of data is being used.

Open-source implementation enables broader adoption: By building on open-sourced agentic frameworks rather than proprietary systems, the approach potentially makes responsible data governance more accessible to organizations of different sizes, supporting wider compliance with privacy regulations.