\n\n
Key developments: Companies are moving away from generic, one‑size‑fits‑all AI models toward a mix of frontier, open and specialised models, but the real focus is now on how those models are governed and deployed. On October 5, 2026 Collibra bought Trail ML to become an AI control‑plane, while Cohere launched North 2, adding quotas, rate limits and a new "Harness" for agent management. Both moves address the growing risk that agents act without proper context—68% of firms have blamed missing or inconsistent business context for erroneous outputs—and the top‑ranked concerns of security, permissioning and vendor lock‑in.
Why it matters: As AI embeds deeper into core processes, enterprises need tools that enforce policies, track usage and retain context across sessions. The new governance layers from Collibra and Cohere, combined with emerging hybrid control‑plane expectations (53% of firms anticipate this by year‑end), aim to close the gap between experimental pilots and production‑ready agents. Microsoft’s Work IQ preview shows the trend of wiring agents directly to live business data, while SAP’s acquisition of TechWolf adds a context‑graph layer to its HR and SuccessFactors suite, further tightening the data‑grounding of AI actions.
What to watch next:
- Adoption of hybrid control planes and their impact on agent orchestration standards.
- Oracle’s Fusion Claw integration, which may set a precedent for embedding agents directly inside ERP systems.
- Continued evolution of security and permission frameworks as 37% of enterprises cite these as the top risk for provider‑controlled orchestration.
- How vendors balance the need for open, specialised models with the governance overhead required for reliable enterprise deployments. [[1, 2, 3]]