Microsoft Work IQ gives enterprise agents events, memory and 60 SharePoint actions
Microsoft has expanded Work IQ with event-driven agent triggers, persistent organisational memory, business-application context and more than 60 native SharePoint actions. Administrators can set tenant-wide action ceilings enforced outside the model, while sanitisation and planned provenance signals aim to reduce prompt-injection risks as Copilot and custom agents move from retrieval into governed execution.
ADNOC adopts OpenAI Frontier to connect decisions across its energy business
ADNOC, OpenAI and G42’s Inception42 have signed a multi-year agreement to connect frontier AI agents across the energy company’s operations. The first work will target field master planning, drawing on decades of reservoir, well and surface-facility data. Inference will run in the UAE, while operational and commercial decisions remain organisational responsibilities.
NVIDIA NemoClaw outlines a controlled stack for specialised enterprise agents
NVIDIA has published a NemoClaw architecture for building specialised agents on infrastructure organisations control. The design combines models, a harness, domain tools, policy-enforcing runtime, optimisation and observability, aiming to give production teams clearer boundaries around data access and actions. The release frames agent quality as a full-stack operating problem rather than a prompt alone.
MoEngage turns more than 400 private APIs into agent tools with Amazon Bedrock AgentCore
AWS has detailed how MoEngage used Amazon Bedrock AgentCore Gateway to expose more than 400 private APIs as governed agent tools. The architecture avoids public internet access, keeps identity and authorisation outside the model and limits context overload by presenting relevant tools on demand, offering a practical pattern for enterprise agents over large internal API estates.
OpenClaw Enterprise opens its agent control plane for internal pilots
OpenClaw announced an open-source Enterprise control plane on 29 September for persistent agents in sensitive environments. It adds multi-tenancy, governance and security boundaries while keeping core components replaceable. The project is still being developed ahead of version 1.0, and OpenClaw recommends internal pilot workloads rather than treating the current release as finished enterprise infrastructure.