Burr is a Python framework for building decision-making applications like chatbots, agents, and simulations. It targets developers needing to manage complex state, track decisions, and integrate LLMs, offering a real-time UI for monitoring and tracing execution.

How It Works

Burr models applications as state machines, allowing developers to define workflows using simple Python functions decorated as actions. It manages state transitions between these actions, enabling idempotent and self-persisting workflows. This approach simplifies complex logic, state management, and debugging, particularly for LLM-driven applications.

Quick Start & Requirements

• Install: pip install "burr[start]"
• Run UI server: burr
• Demo chatbot requires OPENAI_API_KEY environment variable.
• Example: git clone https://github.com/dagworks-inc/burr && cd burr/examples/hello-world-counter && python application.py
• Documentation: Link to documentation
• Video Intro: Link to video

Highlighted Details

• State machine-centric design for explicit workflow management.
• Framework-agnostic, integrating with various LLM and Python libraries.
• Open-source UI for real-time telemetry, monitoring, and debugging.
• Supports non-LLM use cases, including time-series forecasting and hyperparameter tuning.

Maintenance & Community

• Active development with recent commits.
• Discord community available for support: https://discord.gg/6Zy2DwP4f3
• Roadmap includes FastAPI integration, improved error handling, and expanded storage options.

Licensing & Compatibility

• Apache 2.0 License.
• Permissive license suitable for commercial use and integration with closed-source applications.

Limitations & Caveats

Burr does not natively support asynchronous event-based orchestration, unlike some competitors like Temporal. The roadmap indicates planned improvements for asynchronous capabilities and broader framework integrations.