As enterprises move from generative AI pilots to production systems, the hardest problem is increasingly not model selection but controlling the data those models and AI agents can access. Data Dynamics is targeting that gap with Enterprise 2.0.5, a data-management platform update announced at LEAP 2026 that combines model portability, dynamic access controls, compliance evidence and data lifecycle management in a single governance layer.
The enterprise AI stack is becoming more complicated at exactly the point organizations need it to become more controlled.
Large language models can now be deployed through public cloud APIs, private infrastructure, sovereign environments and increasingly autonomous AI agents. But the data feeding those systems often remains scattered across file systems, repositories, applications and legacy platforms. That creates a governance problem that conventional AI security controls do not always solve: knowing what data an AI system can access, why it can access it and whether that access can later be demonstrated to an auditor.
Data Dynamics is positioning its latest release, Enterprise 2.0.5, around that problem.
Announced as generally available at LEAP 2026 in Riyadh, the platform introduces a “Bring Your Own Model” (BYOM) architecture that allows enterprises to run different AI models within their existing trust boundaries. The company says the release supports cloud, on-premises, sovereign and air-gapped deployments.
The distinction matters as enterprises become less willing to build governance around a single model provider.
Organizations may use models from Microsoft, Google or Amazon alongside specialized models and internally developed systems. In some industries, data residency requirements may also prevent sensitive information from being sent to an external AI service. A BYOM approach can therefore give IT teams more flexibility over where inference occurs without separating AI governance from the underlying data.
Enterprise 2.0.5 also applies content-aware and time-bound entitlements to people, AI models and agents. Rather than treating access as a permanent permission, the system is designed around grants that can expire.
That is particularly relevant to the emergence of AI agents.
A conventional employee may request access to a document repository and use that permission for a defined business task. An autonomous agent can potentially perform actions continuously, across multiple systems and without the same human interaction at every step. Persistent credentials or broad permissions consequently become a larger risk.
Time-limited access does not eliminate that risk, but it changes the control model. Instead of asking only whether an agent is authorized, organizations can also define how long it should remain authorized and what content it should be able to reach.
The release also includes an on-demand compliance audit capability intended to generate evidence of governance controls quickly. That moves compliance closer to an operational activity rather than a retrospective exercise.
This is becoming an important distinction for enterprise AI.
Gartner reported in 2025 that 63% of organizations either lacked or were unsure whether they had appropriate data-management practices for AI, and predicted that 60% of AI projects unsupported by AI-ready data would be abandoned through 2026. Gartner’s 2026 research has reinforced the direction of travel: organizations with successful AI initiatives invest up to four times more, as a share of revenue, in foundational areas including data quality and governance.
In other words, data governance is moving from an administrative function toward an AI infrastructure requirement.
Enterprise 2.0.5 approaches another part of that equation through data lifecycle management. Data identified as redundant can be retired before it reaches GPU-intensive processing environments. The immediate benefit is governance and data hygiene; the secondary benefit can be lower compute and storage consumption.
That becomes more significant as enterprises build retrieval-augmented generation systems and agentic applications that continuously retrieve information from corporate repositories.
The regulatory environment adds another layer of urgency.
Saudi Arabia provides a useful example because the country’s Personal Data Protection Law (PDPL) is enforced by the Saudi Data and Artificial Intelligence Authority (SDAIA), an organization that also has a national role in artificial intelligence. SDAIA’s committees reported 48 PDPL enforcement decisions during the first year of active enforcement, with violations including unlawful processing, unauthorized disclosure, inadequate safeguards and marketing without consent.
For enterprises operating AI systems in the Kingdom, that convergence makes data governance difficult to treat as a separate issue from AI governance.
The broader technology market is moving in the same direction. NVIDIA is supplying the infrastructure for increasingly demanding AI workloads, while cloud platforms from Microsoft, Google and Amazon provide access to increasingly diverse model and inference capabilities. Salesforce and Adobe are embedding generative AI into enterprise applications. Across those ecosystems, however, the quality and governance of the underlying enterprise data remain fundamental constraints.
Data Dynamics’ argument is that governance therefore needs to sit closer to the data itself rather than being bolted onto individual AI applications.
That approach also differentiates Enterprise 2.0.5 from model-centric AI platforms. Model providers primarily compete on capabilities such as reasoning, multimodal processing, inference performance and agent frameworks. Data-governance platforms instead compete around discovery, classification, authorization, lineage, compliance and policy enforcement.
The two categories increasingly overlap, but they solve different problems.
For enterprise technology teams, the practical question is not whether Enterprise 2.0.5 can replace an AI platform from Microsoft, Google, Amazon or another provider. It is whether organizations can use those different models while maintaining a consistent policy layer across them.
That becomes especially important for companies operating across jurisdictions, managing sensitive information or deploying AI in sovereign and air-gapped environments.
The release suggests that the next phase of enterprise AI infrastructure will be less about choosing a single “best” model and more about creating a controlled environment in which multiple models, agents and data sources can operate safely.
As AI becomes embedded deeper into enterprise workflows, the data layer may ultimately become the most important place to enforce those boundaries.
Market Landscape
The enterprise AI market is shifting from model experimentation to governed production. Early deployments could often be isolated proofs of concept, but production AI introduces persistent questions around data provenance, access rights, residency, security, retention and auditability.
Gartner’s 2026 research says organizations with successful AI initiatives invest up to four times more in foundational data and analytics capabilities than organizations reporting weaker outcomes. It also found that only 39% of technology leaders were confident their current AI investments would positively affect financial performance.
That creates an opening for a broader category of AI-ready data infrastructure.
Data Dynamics is competing indirectly with several layers of the enterprise technology stack rather than with LLM vendors alone. Data catalogs and governance platforms, identity and access-management systems, cloud security tools, AI governance products and data lifecycle platforms can all address portions of the same problem.
The strategic differentiator for Enterprise 2.0.5 is its attempt to unify those controls around enterprise content and extend them to AI agents and models, while retaining deployment flexibility across cloud, on-premises, sovereign and air-gapped environments.
For CIOs and chief data officers, that architecture could be particularly relevant where regulatory requirements or sensitive workloads make public-cloud-only AI deployments impractical.
Top Insights
- Data Dynamics launched Enterprise 2.0.5 with BYOM architecture, dynamic entitlements and compliance auditing aimed at enterprises scaling governed AI and autonomous agents.
- Time-bound access controls extend conventional data governance to AI models and agents, reducing reliance on persistent permissions for automated enterprise workflows.
- Sovereign and air-gapped deployments address organizations that cannot place sensitive AI workloads entirely within public cloud infrastructure or external model environments.
- Data lifecycle controls can remove redundant information before GPU processing, linking AI governance with infrastructure efficiency, storage optimization and FinOps.
- Enterprise AI governance is becoming infrastructure as regulators, CIOs and data leaders increasingly require demonstrable control over models, agents and underlying data.
Power Tomorrow’s Intelligence — Build It with TechEdgeAI
