Skip to content

Latest commit

 

History

History

arxiv_researcher_agent_with_memori

README.md

Demo

arXiv Researcher Agent Demo with OpenAI Agents & Memori

An advanced AI research assistant built using OpenAI Agents and Memori memory integration. This demo showcases how to create agents that remember research sessions, build upon previous findings, and provide persistent, organized research support focused on arXiv papers.

Features

Agent Capabilities

  • Research Agent: Conducts arXiv research paper searches, generates comprehensive reports, and saves findings to persistent memory
  • Memory Assistant Agent: Retrieves, summarizes, and organizes previous research sessions for easy recall

Memori Memory Integration

  • Persistent Memory: All research sessions are stored and can be referenced in future interactions
  • Conscious Ingestion: Automatically identifies and stores important research information
  • Memory Search: Enables agents to search and build upon previous research

arXiv Research Integration

  • Tavily Search Tool: Advanced search for academic research papers and arXiv publications
  • Academic Focus: Specialized in finding scholarly sources and research papers
  • Fact-Focused Reports: Ensures all findings are verifiable and properly cited

Streamlit UI

  • Interactive Web App: Easy-to-use interface for research and memory queries
  • Chat Modes: Switch between research and memory assistant agents
  • History & Memory Management: View, search, and clear research history

1. Prerequisites

# Install Python dependencies
uv sync

2. API Keys Required

Nebius API Key

Tavily API Key

3. Environment Setup

Create a .env file in this directory:

NEBIUS_API_KEY="your nebius-api-key-here"  # Replace with your actual Nebius AI API key
TAVILY_API_KEY="your-tavily-api-key-here"  # Replace with your actual Tavily API key
EXAMPLE_MODEL_NAME ="moonshotai/Kimi-K2-Instruct" # Replace with any other supported models
EXAMPLE_BASE_URL= "https://api.studio.nebius.ai/v1"

Usage

1. Run the Application

streamlit run app.py

2. Research & Memory Features

  • Research Chat: Ask the agent to research any topic. It will search memory, find relevant arXiv papers, and generate a professional report.
  • Memory Chat: Query your research history, recall previous topics, and get organized summaries of past findings.
  • History Management: View all research sessions or clear memory from the sidebar.

📋 Example Interactions

Research Session

Input: "Research the latest breakthroughs in quantum computing."

The agent will:
1. Search its memory for previous quantum computing research
2. Use Tavily to find relevant arXiv papers and academic sources
3. Cross-reference sources and generate a comprehensive report
4. Save the session to Memori for future reference

Memory Query

Input: "Summarize my research history on AI ethics."

The memory assistant will:
- Search all past research sessions related to AI ethics
- Organize findings chronologically or by topic
- Provide a clear summary and highlight key connections

Research Focus

This agent is specifically designed for academic research and arXiv papers:

  • Academic Sources: Focuses on scholarly papers, research publications, and academic sources
  • arXiv Integration: Specialized in finding and analyzing arXiv research papers
  • Research Methodology: Provides structured analysis of research questions, methods, and findings
  • Academic Standards: Follows proper citation and academic writing conventions