AI News Nuggets

AI scale depends on governed agents and measured inference

Workday has agreed to buy C1.ai to extend identity governance from people to agents, while Microsoft says its Maia 200 inference system now serves Copilot and outlines a larger 2027 rollout. Both announcements put concrete operating dependencies behind AI ambitions, with future integrations and capacity claims still to prove.

Editorial read

This edition collects 2 notes across 2 topic areas and 2 sources. Start with Agent access should follow a named work lifecycle, with permissions removed when a role or task ends, Inference capacity is becoming a design decision for enterprise AI workloads, not just a model-selection detail to get the week's main practical signal before scanning the remaining links.

Edition signal

The October 15 signal is to test who can act and what can serve the workload before scaling enterprise AI

Workday's signed agreement to acquire C1.ai points toward joining agent registration to access provisioning and removal across enterprise applications, but the transaction has not closed and the proposed integration is not a delivered control. Microsoft says Maia 200 is serving Copilot today and plans Maia 300 production deployments and multi-gigawatt capacity from 2027. Its efficiency figures are vendor projections. For an enterprise, the useful test is twofold: can each agent's authority be traced and revoked as work changes, and can the chosen inference path meet actual latency, throughput, cost, residency, and portability requirements?

SecurityInfrastructure

Security October 15 note

Official Workday release

Agent access should follow a named work lifecycle, with permissions removed when a role or task ends

Source: Workday

Workday has signed a definitive agreement to acquire C1.ai, with a close expected in the fourth quarter of its fiscal 2027, subject to customary conditions. Workday says its existing Agent System of Record provides a place to register and manage agents. The proposed C1.ai combination would connect that inventory to access decisions across Workday and other enterprise applications, using events such as a person joining, changing role, or leaving. The acquisition and cross-application integration are plans, not current delivered capabilities.

Why this matters: Treat agent identity as an access lifecycle rather than an inventory field. Record the human or service owner, delegated purpose, target systems, allowed actions, approval path, and expiry for each agent. Test how access changes when the owner moves, a task ends, or an agent is retired, and confirm that revocation propagates across connected applications. Evaluate Workday's and C1.ai's current capabilities separately until the deal closes and the proposed integration can be tested in a real environment.

Read Workday's C1.ai acquisition announcement

Infrastructure October 15 note

Official Microsoft release

Inference capacity is becoming a design decision for enterprise AI workloads, not just a model-selection detail

Source: Microsoft

Microsoft says Azure Maia 200 is already serving production workloads including Microsoft Copilot, with GitHub Copilot and Microsoft Foundry planned to follow by the end of 2026. It plans to begin Maia 300 production deployments and expand the Maia family to multi-gigawatt scale in 2027. Maia 300 is designed for long-context, latency-sensitive, and multi-step agent workloads; Microsoft's throughput and efficiency comparisons are its own projections, not independent production benchmarks for customer workloads.

Why this matters: As agents make longer and more frequent model calls, compare the full serving path on a representative workload: latency at the required concurrency, tokens per unit of cost and power, model choice, data location, and failure recovery. Ask which current service actually runs on Maia and which benefits depend on a future generation. Treat the 2027 capacity plan as a planning signal, then base procurement and architecture decisions on observed service behavior and contract terms when available.

Read Microsoft's Azure Maia announcement