According to recent DBTA research, 59% of organizations are actively piloting or scaling AI initiatives across the enterprise. Yet, fewer than half of these initiatives are succeeding for most organizations. Data quality, data access, trust, and data lineage remain major challenges, contributing to 77% of AI failures today.
To address these challenges, organizations are increasing investments in data quality, governance, and compliance to strengthen the data foundation for AI. In parallel, many are prioritizing stronger security controls and safeguards to better protect sensitive data across AI environments.
DBTA recently held a roundtable webinar, Trusted AI at Scale: Modern Data Governance and Security Strategies, with experts who discussed modern governance and security strategies that support trusted AI at scale.
According to Dany Nehme, director, solutions engineering at insightsoftware, multi-MCP AI creates compliance gaps. Every extra MCP is a gap in your audit trail. Cross-source AI inference is invisible to audit systems. And multi-MCP architectures cost more, run slower, and produce less accurate answers.
Simba Intelligence solves this at the access point, Nehme said. The platform offers one governed semantic layer, one MCP. Every query is logged, attributed, and policy-enforced.
The consumer journey is fragmented across siloed systems …so the AI agent works off duplicate, inconsistent core data. Without the necessary policy guardrails, explained Gaurav Gera, director, product management at Reltio. The AI agents’ poor decisions quickly cause frustrations and attrition.
Without data unification, AI lacks trusted core data. Missing unified context means AI agents are fast, confident, and wrong, Gera said. A real-time system of context solves this problem.
Key requirements for a system of context include the following:
- Context-rich knowledge graph with entities, metadata, relationships.
- Open, interoperable platform with MCP and API-first architecture.
- Real-time, event-driven performance in milliseconds at scale.
- No-code deployment of unified, agent-ready data products at scale.
- Governed by design: integrations inherit the platform’s security, RBAC permissions, lineage, and audit policies—for compliance and transparency across every flow.
Taryn Stebbins, community of practice leader at Informatica from Salesforce, noted that there is a growing gap between ambitions for AI and the necessary level of data maturity.
To turn raw data into trusted data, a successful platform must offer the following:
- Measurable data quality
- Technical and business metadata
- Automated lineage
- Proactive governance
AI is shifting the software delivery bottleneck, according to Woody Evans, vice president, global sales engineering at Perforce Delphix. AI velocity without data readiness creates delivery chaos.
Governance and security strategies support trusted AI, Evans concluded.
The foundation of trusted data environments includes:
- Maintain Data Shape: Let the use case drive the solution that best maintain shape and security.
- Mitigate Sovereignty Threat: Protection, policy, access controls, and audit are built in.
- Join Lineage and Velocity: Automated data operations at AI pace with built in lineage and auditability.
For the full webinar, featuring a more in-depth discussion, Q&A, and more, you can view an archived version of the webinar here.