FloAI (flo-ai) is an open-source Python framework developed by Rootflo that enables developers to build production-ready AI agents and multi-agent teams through YAML-first configuration. It is positioned as an alternative to LangGraph and CrewAI, offering a middle ground between ease-of-use and deep control, similar to how Keras relates to TensorFlow. The framework is available on PyPI as the 'flo-ai' package and on GitHub under the rootflo organization. The core design philosophy is composability: complex AI systems are assembled from smaller, reusable components. Agents and teams can be described entirely in YAML, making it approachable for teams who want readable, version-controlled agent architectures without writing extensive Python boilerplate. The framework supports multiple LLM providers including OpenAI, Anthropic, Google, and Ollama through a consistent interface. FloAI is part of the broader Rootflo ecosystem. Rootflo also develops Wavefront, an open-source enterprise AI middleware platform targeting BFSI (Banking, Financial Services, and Insurance) use cases such as contact center automation, underwriting workflows, and compliance automation. Wavefront is a separate but related product that acts as a connective layer between frontend applications and backend data pipelines, with a roadmap targeting GA release in Q1 2026. The framework includes built-in observability tooling for production deployments, LangChain tool compatibility, a @flo_tool decorator for easy custom tool creation, and support for RAG (Retrieval-Augmented Generation). A React-based visual studio interface is also part of the Wavefront platform for workflow design. The project is under active development and APIs may change; the Wavefront middleware component was in pre-release as of late 2025. Key features: - YAML-first agent and team configuration for readable, version-controlled architectures - Composable component model: build complex multi-agent systems from reusable building blocks - Multi-LLM provider support (OpenAI, Anthropic, Google, Ollama) via consistent interface - LangChain tool compatibility and @flo_tool decorator for custom tool creation - Built-in observability for production deployments - RAG (Retrieval-Augmented Generation) support built in - Wavefront enterprise middleware layer for BFSI workflows (contact center, underwriting, compliance) - React-based visual studio interface for workflow design
Open-source and free (PyPI package). Rootflo Wavefront Cloud multi-tenant offering planned for Q1 2026; pricing not yet public.