ultrawork

To execute independent tasks concurrently, you can fire multiple tasks in parallel across executor agents. This maximizes throughput and reduces total wall time for workloads like batch model inferences or data processing.

Parallel task orchestration is the process of routing and executing independent tasks concurrently across multiple agents. You need it when running workloads like batch model inferences to reduce total wall time and maximize throughput.

Yes, you can run long-running operations in the background. The execution lifecycle is managed by omx_state, allowing you to fire parallel tasks and maintain background operations without persistence concerns.

Yes, the workflow supports explicit model parameter passing. This allows you to route tasks to appropriate agent tiers and manage parallelism while retaining responsibility for the final results.

Without persistence, you must handle final results manually as you retain responsibility for them. However, you can use it with the Ralph/autopilot stack to layer persistence and manage the full execution lifecycle.

Task independence classification is supported within the workflow to ensure tasks can be executed concurrently. You must verify tasks have no dependencies before firing them to executor agents to maximize throughput.