Build, discover, and monetize AI agents.

Oracle adds four “Fusion Agentic Applications” for supply chain

What changed: Oracle announced four new Fusion Agentic Applications (Inventory Planning Command Center, Supplier Qualification Workspace, Production Readiness Workspace, and Kanban Administrative Workspace) and an AI Agent Studio to build, connect, and run reusable agents inside Oracle Fusion Cloud SCM.

Why it matters: These are not assistant widgets — Oracle positions the apps to autonomously progress routine supply‑chain work inside existing enterprise controls, which lets procurement, planning, and manufacturing teams automate recurring decisions without rebuilding core systems. That reduces time-to-action for disruptions and shifts some operational load off busy teams.

Try/watch: If you run ERP/SCM ops, evaluate whether the apps can integrate with your existing workflows and SLAs in a non-disruptive way; pilot inventory and supplier workflows first to measure exception rates and human escalation points.

AvePoint: enterprises report widening visibility and security gaps as agents scale

What changed: AvePoint published its 2026 State of AI report finding 46.9% of employees use AI agents weekly or daily and that 88.4% of organizations experienced at least one agent‑related security incident in the past year. The study also reports a sharp rise in unsanctioned agent use and widespread deployment delays due to governance concerns.

Why it matters: Those numbers mean adoption is outpacing basic operational controls — buyers and operators should treat agent observability, identity, and data lineage as first‑order features, not optional add‑ons, because breaches and blind spots show up quickly once agents begin taking action across systems.

Try/watch: Run an audit of agent access and data use (who’s building agents, what data they touch, where outputs are stored) and prioritize short pilots for agent management tooling or agent inventories before scaling production usage.

Trust3 AI connects its agent governance stack to Microsoft Copilot Studio

What changed: Trust3 AI announced an integration that lets security and platform teams discover Copilot Studio agents (including shadow agents), capture execution history, enforce runtime guardrails, and stop agents in real time. The release is available now and will be demoed at AI Engineer World’s Fair.

Why it matters: As vendors ship agent‑building surfaces, the control problem moves from theoretical to practical: teams need continuous discovery, tamper‑evident logs, and the ability to kill misbehaving agents. This integration is a practical example of how operators can add governance without blocking developer productivity.

Try/watch: If you use Copilot Studio or similar agent platforms, test agent discovery and kill‑switch workflows in a staging environment and validate that identity and audit trails map back to human owners before granting agents broad data or action privileges.