Enterprise AI is now sold on governance first, models second

The partnership with governance in the name

OpenAI and Databricks formed a partnership in September 2025, and the partnership between OpenAI and Databricks is a multi-year agreement. OpenAI is one of Databricks' largest customers, and Databricks engineers are receiving access to OpenAI Codex. That last detail is easy to miss: the partnership runs both ways, with Databricks using the models it sells access to, and getting early access to new OpenAI models in return. m:…9303

The product is the governance, not the model. Databricks serves OpenAI models next to company data under Unity Catalog governance, Genie Ontology is the context layer in Databricks that stores company-approved metrics and business definitions, and the Databricks Genie plugin allows users to query company data within ChatGPT. The pitch is that the model answers from your governed data, using your approved definitions, rather than from whatever it remembers. m:…9303

The controls are concrete. The Databricks Unity Gateway allows companies to manage the rollout of OpenAI Codex to engineers, logs every OpenAI Codex request, and lets engineering leaders set budgets and review spend by team. No external API key is needed when Databricks serves the model. That is the whole enterprise AI proposition in one sentence: the model is rented, the data stays governed, every request is logged, and the bill is visible by team. m:…9303

Microsoft's controls, extended outward

Airia's pitch starts from a gap in Microsoft's own stack. Agent 365 became generally available for commercial customers on May 1, 2026, and Agent 365 works best with Microsoft E5 as a prerequisite. But Microsoft Purview's classic DSPM for AI does not support Agent 365, and content created by Agent 365 agents does not inherit sensitivity labels from source files. In other words, the agent platform shipped before the governance for it did, and the labels that protect your documents do not follow the agent's output. m:…5942

Airia acts as an external threat detection provider for Copilot Studio, integrates with Microsoft Sentinel to stream agent activity and policy violations, and provides a routing engine for model access with budget controls. Agent 365 uses Entra for identity and risk-based access controls. Airia can run as shared SaaS or as a private deployment in an Azure tenant, and is transactable on Azure Marketplace and counts toward Microsoft spend commitments. The commercial detail matters: buying governance through the marketplace counts against the same commitment as buying the models, so the control plane is not a separate budget fight. m:…5942

Observability as the third leg

Dynatrace is integrating its observability context into Microsoft workflows and natively ingesting telemetry from Azure environments, extending OpenTelemetry and OpenInference instrumentation to applications built on Azure AI Foundry. Its Model Context Protocol server connects observability metrics and runtime insights to external AI agents like GitHub Copilot, which means the agent can see the system it is working on, not just the code. m:…4172

Dynatrace AI observability for Azure AI Foundry captures traces, prompt execution signals, token usage, latency, and cost metrics, and the Azure SRE Agent uses real-time dependency graphs to evaluate operational issues across Azure services, apps, and infrastructure. The SRE agent is the point where observability becomes action: the system that watches the AI can also fix what it finds. Microsoft Ignite 2026 will take place November 17 to 20, 2026, at the Moscone Center in San Francisco, where Dynatrace is showcasing these integrations, and Dynatrace is hosting a virtual Lunch and Learn on December 10, 2026, regarding observability on Microsoft Azure. m:…4172

Three vendors, one pattern. Databricks serves OpenAI models next to company data under Unity Catalog governance. Airia provides a routing engine for model access with budget controls. Dynatrace captures traces, prompt execution signals, token usage, latency, and cost metrics. The model is the commodity. The context layer, the routing engine, and the observability are where the enterprise value now sits, and where the next buying decisions will be made. m:…9303 m:…5942 m:…4172