Announcing Workflows 1.0: A Lightweight Framework for Agentic systems

LlamaIndex Workflows 1.0 has been officially released as a standalone framework designed to facilitate the creation of complex, multi-step AI applications in Python and Typescript, offering developers improved control over application logic. Initially part of llama_index, Workflows has evolved into its own package, enabling the orchestration of AI models, APIs, and decision-making processes across various use cases, including document processing pipelines, multimodal AI applications, and customer support automation. This release emphasizes an async-first, event-driven architecture, making it suitable for routing between different capabilities, implementing parallel processing, and maintaining state across multiple steps. The new version introduces features such as typed workflow state, dynamic resource injection, and optional observability integrations, while still being compatible with existing llama_index functionalities. This development highlights Workflows' independence, encouraging broader adoption and contributions beyond the original llama_index ecosystem.