Welcome to your monthly watsonx Orchestrate newsletter!

This month we're focusing on one of the most important developments in enterprise AI operations: the watsonx Orchestrate Agent Control Plane. As organizations move from experimenting with individual AI agents to managing entire agent ecosystems, the need for visibility, governance, and operational oversight becomes critical. 🚀

TL;DR (2‑minute scan for builders)

Agent Control Plane
→ Centralized visibility, governance, and optimization for AI agents across platforms
→ Why it matters: Reduce agent sprawl and scale AI operations with confidence
AgentOps & July Release Highlights
→ AgentOps (AWS Preview), Agent Optimization Assistant, A2A agent management, and workflow notifications
→ Why it matters: Build, evaluate, and operate agents more effectively
ADK 2.13.0 Updates
→Agent Skills, enhanced trace observability, expanded workflow callbacks, and GPT‑5.4 support
→Why it matters: Accelerate development of more capable and manageable AI agents
Upcoming webinars
→ Live demos + deeper dives into real use case

Featured Highlights

Agent Control Plane: Bringing Order to Enterprise Agent Ecosystems

As enterprises adopt more AI agents, a common challenge emerges: how do you effectively manage, monitor, and govern agents operating across different business processes, cloud environments, and AI frameworks?

The watsonx Orchestrate Agent Control Plane provides a centralized operational layer that helps organizations observe, govern, and optimize AI agents across the enterprise, regardless of where those agents were built or where they run.

Key capabilities include:

• Centralized visibility into agent activity and performance
• Governance through policies, controls, and guardrails
• End-to-end monitoring and traceability
• Cost and usage optimization insights
• Enterprise-grade access control and security management
• Support for managing agents across multiple frameworks and platforms

Many organizations are rapidly expanding beyond a handful of use cases into dozens or even hundreds of AI-powered workflows. Without a centralized operational model, teams can struggle with visibility, compliance, performance management, and cost control

Learn more about the Agent Control Plane

Control Plane in Action

A great example of this capability was recently highlighted by IBM's Andreas Schneider, demonstrating how the watsonx Orchestrate Control Plane can monitor and govern external agents, including agents running on Google Vertex AI. The example showcases how organizations can maintain centralized oversight while supporting heterogeneous AI environments.

Use this capability when you need to:

• Monitor agents across multiple platforms
• Improve governance and compliance
• Understand agent decision-making through traceability
• Optimize costs and operational performance
• Establish enterprise-wide standards for AI operations

See a real-world example in action

What's New

July Mid-Month Release Highlights

Rather than re-listing every release item here, we encourage you to review IBM's detailed release content for the complete walkthrough of new capabilities. This release introduces several significant enhancements across agent development, workflow automation, connections management, analytics, and governance.

Notable additions include:

• AgentOps (Preview on AWS) for evaluating and optimizing agents
• Agent Optimization Assistant for AI-assisted agent improvement
• Enhanced management of external A2A agents
• Workflow runtime notifications to external systems
• Improved controls management and governance capabilities
• Additional workflow and administration enhancements

Dive into the latest release updates

New in the ADK: Smarter Workflows, Parallel Processing, and Multi-Language Support

The watsonx Orchestrate ADK 2.13.0 release introduces new Agent Skills for packaging and reusing specialized agent capabilities, enhanced trace observability for easier debugging, expanded workflow callback events, support for document extraction by page range, and the ability to suppress agent summarization when workflows complete. The release also adds support for premier models such as GPT-5.4 and continues the move toward the recommended ReAct Core agent style for building more capable AI agents.

Explore the full ADK 2.13.0 release notes

As always, thank you for your feedback, collaboration, and continued support.

We'd love to hear from you: How are you planning to operationalize and govern AI agents within your organization, and what topics would you like us to cover in a future Technical Touchpoint?