What is Blueprints?
Blueprints is a developer-first hub for open-source AI workflows, designed to stop searching and start building. It offers a curated collection of reusable, modular blueprints that demonstrate practical implementations of AI tasks using open-source models and tools. Each blueprint includes a README, pyproject.toml, organized folder structure, and a demo app built with Streamlit or Gradio, or a Jupyter notebook. They feature config files for easy customization, CLI support, CPU compatibility for most local setups, Google Colab notebook options, PyPI package availability, and Dockerfiles for the demo app. Diagrams and setup guidance docs using mkdocs are also included.
Key blueprints cover a range of use cases: running AI agents in the browser via WebAssembly (WASM) using Pyodide and the OpenAI Agents Python SDK; re-adjusting Whisper models to user-predefined texts for improved transcription accuracy of uncommon terms; parsing structured data from scanned documents using open-source OCR and LLMs; building task-specific agents and comparing agent frameworks for specific use cases like finding a surfing spot; training models to detect synthetic and modified audio content; transcribing audio files with open-source Whisper models via Speaches; converting unstructured documents to Markdown using Docling CLI; personalizing local timeline algorithms that users can customize; and finetuning LLMs with federated AI using Flower for collaboration on private datasets without sharing raw data.
Blueprints emphasizes open-source models and tools, ensuring transparency and flexibility. They are ideal for developers, researchers, and AI practitioners who want to quickly prototype, customize, and deploy AI solutions locally or in the cloud. The hub encourages community contributions and feedback to improve the experience. By providing ready-to-run workflows, Blueprints accelerates development, reduces duplication of effort, and enables learning through hands-on examples. Whether you're building a transcription system, document parser, or agent framework, Blueprints offers a solid foundation to bring your ideas to life. Visit mozilla.ai for the latest updates and to share your blueprint ideas.
Use Cases
Key Features
curated AI blueprints modular reusable workflows open source models Streamlit/Gradio demos CLI support CPU compatibility Google Colab integration Dockerfile included config file customization mkdocs documentationWho is it for?
AI developers, machine learning engineers, data scientists, open, source enthusiasts, software engineers, researchers, hobbyists
