← All projects

Telegram GitHub project

MindPulse-memory-bot

A professional Telegram AI assistant with long-term Pinecone memory and semantic search.

Problem

Typical AI assistants lose important context between conversations, forcing users to repeat relevant information.

Solution

Implemented short-term and long-term memory using embeddings, Pinecone, and semantic retrieval of relevant context before each LLM request.

Result

The assistant can reuse relevant context from previous interactions, reducing repeated input and creating a foundation for more consistent customer and internal AI workflows.

Technologies

Python