> For the complete documentation index, see [llms.txt](https://yoyo-11.gitbook.io/agent/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://yoyo-11.gitbook.io/agent/technical-architecture.md).

# Technical Architecture

YoYo is built on an  AI Agent architecture + RAG (Retrieval-Augmented Generation)  and the  LangChain  framework, enabling human-like perception → reasoning → action.  

####  Key Modules :  

1\.  **Perception** : Processes user inputs (text/voice).  

2\.  **Reasoning** : Leverages LLMs to analyze queries, decide actions (e.g., external retrieval), or fetch answers from internal knowledge.  

3\.  **Action** : Executes tasks (RAG retrieval, database/API calls, real-time crypto pricing).  

4\.  **Memory** : Stores conversation history, task status, tone, and user feedback.  

####  RAG Workflow :  

1\.  **Retrieval** : Searches external sources for relevant info.  

2\.  **Augmentation** : Contextualizes retrieved data for LLMs.  

3\.  **Generation** : Produces responses aligned with YoYo’s personality and tone.  

####  Advanced Architecture :  

* LangGraph-based agents  outperform traditional prompt/workflow systems by enabling "cyclic thinking" and self-iteration through human feedback.
* Multi-AI Scoring : Inspired by DeepSeek’s R1 model, outputs are filtered by scoring agents for professionalism and tone consistency. 

 **Future Upgrade** : Cross-platform long-term memory will sync data (Twitter, Telegram) to maintain consistent user relationships.