https://github.com/rdhillbb/ozzss
Agentic application built on Langchain and DSPy
https://github.com/rdhillbb/ozzss
cloud dspy genai-chatbot golang groq langchain llama3 openai python3
Last synced: 6 days ago
JSON representation
Agentic application built on Langchain and DSPy
- Host: GitHub
- URL: https://github.com/rdhillbb/ozzss
- Owner: rdhillbb
- Created: 2024-06-02T22:32:31.000Z (12 months ago)
- Default Branch: main
- Last Pushed: 2024-06-10T05:45:25.000Z (11 months ago)
- Last Synced: 2025-04-30T02:40:12.632Z (20 days ago)
- Topics: cloud, dspy, genai-chatbot, golang, groq, langchain, llama3, openai, python3
- Language: Python
- Homepage:
- Size: 70.3 KB
- Stars: 1
- Watchers: 2
- Forks: 1
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# Welcome to Ozzss
The project **Ozzss** is a journey to building a network agentic platform. By the completion of the project, the app **Geppetto** will make use of network agents and network APIs that are registered with the clearing house called Ozzss.
We will not do a big bang release as most projects. Each release will be a set of lego blocks that can be integrated into your application. Ozzss will be built on **Langchain** and **DSpy**. Langchain was selected as we started our initial journey one year ago with Langchain. This is not to say that we feel Langchain is a better platform than **LlamaIndex**. We are just comfortable with Langchain and too lazy to explore another orchestration layer. As we venture along our journey, for sure we will integrate LlamaIndex.
Our position is that what distinguishes one GenAI application from another are the tools, specialized LLMs, and prompts. Most importantly, the system prompt that sets the persona of the application. Switching between one domain to another should be a matter of dynamically loading in the prompts and tools to define the nature of the application. From our experience, we see this can be done 85% of the time, as there will always be corner cases.
Once the controlling LLM app is architected, the definition of the application will be based on the system persona and will interface with Ozzss to locate tools, agents, and LLMs to respond to requests and questions.
---
### Activity
**June 5th Release (LLMS):** LLMObjects - Simplifying Platform and Model Integration
**On Deck:** Framework to integrate tools through a YAML configuration file or loader
---
# Note:
**We welcome comments be it good, bad, ugly, or flames as we do not claim to be know-it-alls. Changes to improve our python code is very much welcomed**