Working with Karli Studio¶

Agentlab is the AI workflow engine behind Karli Studio. While Agentlab handles flow execution, Karli Studio provides the user-facing experience — including the chat interface, file management, and administrative features.

This section covers how Agentlab and Karli Studio interact, from the perspective of a flow designer building workflows for end users.

Architecture Overview¶

flowchart TD
    subgraph Studio["Karli Studio"]
        ChatUI["Chat UI"]
        FileMgmt["File Management"]
        Admin["Admin"]
        Middleware["Middleware"]
        ChatUI --> Middleware
        FileMgmt --> Middleware
        Admin --> Middleware
    end
    Middleware --> Agentlab["Agentlab
(Flow Engine)"]
    Agentlab --> Langfuse["Langfuse
(Observability)"]

User interactions in the Karli Studio chat (messages, file uploads, chip selections, context choices) are routed through the middleware to Agentlab, which executes the appropriate flow and returns results back to the chat interface.

Key Integration Points¶

• File Uploads — how files uploaded in the chat reach your flow components.
• Observability — tracing flow execution with Langfuse.