Burr Framework

Burr is a lightweight Python framework designed to model and execute application logic as state machines represented by action-driven graphs. It's especially well suited for AI agent workflows, simulations, and dynamic decision-making systems. Burr simplifies managing state, complex decisions, and human feedback cycles by representing workflows as graphs with conditional branching and looping. It also includes a self-hostable observability UI for monitoring and debugging execution flows in real time. Burr aims to provide a vendor-neutral, open-source standard for orchestrating fine-grained computational graphs, empowering developers to build scalable and maintainable AI and data workflows while enabling software engineering best practices like testing, modularity, and provenance.

Key Features

• Express your application's logic as state machines using simple Python functions with clear tracking of state changes.
• Comes with a self-hostable, real-time observability UI that helps monitor, trace, and debug workflows effectively.
• Framework-agnostic and lightweight, designed for fine-grained orchestration beyond heavyweight workflow tools like Apache Airflow.
• Supports modular, reusable action building blocks and integrates easily with telemetry and persistance systems.

Use Cases

• Building AI agents and chatbot workflows that require complex decision-making and state persistence.
• Running simulations or interactive applications where state management and event-driven logic are essential.
• Managing scalable, observable business logic and data workflows with conditional branching and looping.

Technical Specifications

• Fully written in Python with minimal external dependencies to keep it lightweight and easy to integrate.
• Supports remote or local execution with optional Docker deployment for hosting the observability UI.
• Provides extensible APIs for defining actions, states, transitions, and integrates with OpenTelemetry for metrics and tracing.