workflow-engine

You can orchestrate AI-powered workflows using a WorkflowBuilder to assemble synchronous and async-batch stages with type-safe schemas. The unified runtime manages pipeline execution, dependencies, and Prisma persistence for jobs, logs, and artifacts.

Tracking AI API costs across pipeline stages is handled via integrated AI helpers that provide hierarchical cost tracking. This monitors expenses across individual stages and throughout the entire suspended and resumed workflow execution.

Yes, you can suspend and resume long-running tasks in a multi-stage workflow. The orchestration runtime supports suspended and resumed executions, allowing pipelines to pause during async-batch stages and continue processing reliably later.

You define type-safe input and output schemas for async-batch stages using defineAsyncBatchStage. This ensures precise execution semantics and data validation across synchronous and asynchronous pipeline operations.

Yes, Prisma persistence supports job logs and artifacts for AI workflow orchestration. The unified runtime leverages Prisma to manage persistence, ensuring data and artifacts are stored securely across suspended and resumed executions.

You need type-safe stage definitions for AI automation pipelines when building multi-stage workflows that require strict input and output schema validation. This reduces complexity and prevents runtime errors across synchronous and async-batch executions.