Ecosyste.ms: Awesome

An open API service indexing awesome lists of open source software.

Awesome Lists | Featured Topics | Projects

https://github.com/dieharders/obrew-studio-server

🍺 Obrew Studio Server: A self-hostable machine learning engine. Build agents and schedule workflows private to you.
https://github.com/dieharders/obrew-studio-server

agents ai assistant chatbot desktop-app inference-engine llamacpp localai offline-llm productivity rag self-hosted text-generation

Last synced: about 23 hours ago
JSON representation

🍺 Obrew Studio Server: A self-hostable machine learning engine. Build agents and schedule workflows private to you.

Awesome Lists containing this project

README 

        
# 🍺 Obrew Studio: Server - Your Personal Ai Engine






## Table of Contents



- [Features](#app-features-roadmap)

- [How to Use](assets/how-to-use.md)

- [Getting Started](assets/getting-started.md)

- [API Documentation](assets/api-docs.md)

- [Build Steps](assets/build-steps.md)

- [Compiling & Packaging](assets/compiling-packaging.md)

- [Deploy & Release](assets/deploy-release.md)



## Introduction



The goal of this project is to be an all-in-one solution for running local Ai that is easy to install, setup and use. It is a native app that runs a server which handles all basic building blocks of building with Ai: inference, vector memory, model file manager, agent builder, GUI.



## How It Works



This backend is a Python server built with FastAPI. A WebUI is provided called [Obrew Studio WebUI](https://studio.openbrewai.com/) to access this server. You can also access it programmatically via the [API](assets/api-docs.md).



Launch the desktop app locally, then navigate your browser to any web app that supports this project's api and start using ai locally with your own private data for free:



## App Features Roadmap



- ✅ Run locally

- ✅ Desktop installers

- ✅ Save chat history

- ✅ CPU & GPU support

- ✅ Windows OS installer

- ❌ MacOS/Linux installer

- ❌ Docker config for cloud/server deployment

- ❌ Production ready: This project is currently under active development, there may be bugs



## Ai Features Roadmap



- ✅ Inference: Run open-source LLM models locally

- ✅ Embeddings: Create vector embeddings from a file/website/media to augment memory

- ✅ Knowledge Base: Search a vector database with Llama Index to retrieve information

- ✅ Agents: Customized LLM, can choose or specify tool use

- ❌ Workflows: Composable automation of tasks, teams of agents, parallel processing, conditional routing

- ❌ Monitors: Source citations, observability, logging, time-travel, transparency

- ❌ Support multi-modal: vision, text, audio, 3d, and beyond

- ❌ Support multi-device memory sharing (i.e. cluster of macs running single large model)

- ❌ Support voice-to-text and text-to-speech

- ❌ Auto Agents: Self-prompting, autonomous agent given tools and access to a sandboxed OS env



## Supported Model Providers



This is a local first project. The ultimate goal is to support many providers via one API.



- ✅ [Open-Source](https://huggingface.co)

- ❌ [Google Gemini](https://gemini.google.com)

- ❌ [OpenAI](https://openai.com/chatgpt)

- ❌ [Anthropic](https://www.anthropic.com)

- ❌ [Mistral AI](https://mistral.ai)

- ❌ [Groq](https://groq.com)



## Learn More



- Backend: [FastAPI](https://fastapi.tiangolo.com/) - learn about FastAPI features and API.

- Inference: [llama-cpp](https://github.com/ggerganov/llama.cpp) and [llama-cpp-python](https://github.com/abetlen/llama-cpp-python) for LLM inference.

- Memory: [Llama-Index](https://github.com/run-llama/llama_index) for data retrieval and [ChromaDB](https://github.com/chroma-core/chroma) for vector database.

- WebUI: Vanilla HTML and [Next.js](https://nextjs.org/) for front-end UI and [Pywebview](https://github.com/r0x0r/pywebview) for rendering the webview.