prose

You orchestrate multi-agent AI workflows by defining agent templates and spawning sessions using declarative programs. This allows tasks to execute in parallel or sequentially while managing state and imports across runs.

Declarative AI agent orchestration manages state and persistence across complex pipelines using filesystem, SQLite, PostgreSQL, or in-context backends. This ensures context passing and state management remain consistent throughout the workflow execution.

Yes, you can use nested imports for AI agent orchestration in production environments. The runtime semantics support building complex production orchestrations and experiments with persistent agents and nested program structures.

The best way to coordinate specialized AI agents on a feature rollout is using a captain's chair pattern. This orchestration method allows a central coordinator to manage and dispatch tasks to specialized agents effectively.

You run a sample multi-agent pipeline to test session orchestration by executing a declarative program file. This initiates the structured VM runtime to orchestrate tasks and demonstrate the workflow mechanics end to end.

Yes, AI agent orchestration works without external databases by using in-context backends or the filesystem for state management. You can also opt for SQLite or PostgreSQL if your production pipeline requires persistent database storage.