ultrawork

You can parallelize independent tasks across multiple executor agents to reduce total execution time and improve throughput by firing them simultaneously instead of sequentially.

Model routing enforces explicit assignment of each parallel task to the correct model tier. You manage delegation while maintaining control over completion, ensuring tasks match the appropriate executor level.

Yes, you can use run_in_background for long operations to improve efficiency in large, multi-agent workloads. This allows continuous orchestration while lengthy parallel tasks execute without blocking.

The best way to orchestrate high-throughput tasks is to delegate independent work in parallel to the appropriate agents, letting the lifecycle components manage execution, verify task completion, and prevent new errors.

Yes, you should read agent tier guidance before your first delegation to ensure each task is routed to the correct model tier, which improves efficiency and alignment within the multi-agent workload.

You should avoid parallel execution when tasks are dependent on each other. This approach applies when several tasks can run concurrently and independently, allowing you to delegate work and manage completion without sequential delays.