Agent Studio in Hybrid Manager

Agent Studio in Hybrid Manager gives you an integrated way to create, manage, and govern AI Assistants running entirely within your Hybrid Manager project environment — fully aligned with your Sovereign AI architecture.

You can use Agent Studio to build Assistants that:

• Interact with models deployed through Model Serving.
• Retrieve content from Knowledge Bases.
• Apply Rulesets for governance and compliance.
• Execute Tools that perform actions or integrate with external systems.
• Maintain Threads for stateful, auditable conversations.

For a deeper background, see the Agent Studio concepts page in the AI Factory hub.

How Agent Studio works in Hybrid Manager

Agent Studio components run entirely in your Hybrid Manager project’s Kubernetes cluster, with full observability and governance:

• Assistants are defined in Agent Studio and run within your project infrastructure.
• Assistants can call models deployed via Model Serving — see Model Serving.
• Assistants retrieve content from Knowledge Bases built within the same project — see Knowledge Bases explained and AI Accelerator Knowledge Base page.
• Assistants apply Rulesets — defined in your Agent Studio configuration — see Rulesets explained.
• Assistants can invoke Tools implemented within your project — see Tools explained.
• Threads are captured automatically to support auditability and review — see Threads explained.

All these components operate under your control:

• Data stays in your infrastructure
• Models run on your GPUs, in your Kubernetes cluster
• Knowledge Bases and embeddings stay in your Postgres and object storage
• Assistant behavior is fully observable and governed through Rulesets

This supports Sovereign AI principles — see Sovereign AI explained.

Typical Agent Studio flow in Hybrid Manager

1. Define a Knowledge Base — How-To: Create a Knowledge Base.
2. Create and apply Rulesets — How-To: Create a Ruleset.
3. Define a Retriever to control Knowledge Base search — How-To: Create a Retriever.
4. Deploy Tools for API calls or business logic — How-To: Create a Tool.
5. Deploy supporting models via Model Serving — How-To: Deploy a NIM container.
6. Create your Assistant in Agent Studio — How-To: Create an Assistant.
7. Test and validate using Threads — How-To: View Threads.

Key links and learning resources

• Agent Studio Hub concepts page
• Create an Assistant — How-To Guide
• Assistants explained
• Rulesets explained
• AI Accelerator Knowledge Bases
• Tools explained
• Threads explained
• Model Serving in Hybrid Manager
• Sovereign AI explained

Next steps: After building your first Assistant, explore advanced topics in the AI Factory Learning Paths.