Skip to content

full-stack research assistant (FastAPI + OpenAI + SerpAPI) with live web search, citation generation, and content filtering to reduce hallucinations.

Notifications You must be signed in to change notification settings

manasa-26/AI-Research-agent

Repository files navigation

🧠 AI-Powered Research Assistant

An AI-powered research assistant that takes user queries, fetches real-time web data, and generates concise, source-cited answers using a language model.

Built with a modern frontend (Vite + React.js) and a modular backend (FastAPI, OpenAI, SerpAPI), this assistant helps users get accurate answers with references—fast.


🔙 Backend – FastAPI, OpenAI, SerpAPI

The backend is a modular, high-performance service built with FastAPI. It handles:

  • 📝 Query Intake: Accepts natural language input from the frontend.
  • 🌐 Web Search Integration: Uses SerpAPI to gather relevant real-time web data.
  • 📑 Information Extraction: Filters and selects the most relevant snippets.
  • 🧠 LLM Summarization: Utilizes OpenAI's GPT models to produce a well-structured answer.
  • 🔗 Citation Handling: Formats output with inline numbered citations, mapped to actual URLs.

Technologies:

  • FastAPI
  • OpenAI GPT
  • SerpAPI
  • Python, Pydantic
  • Uvicorn (async server)

💻 Frontend – Vite + React.js

A clean and responsive single-page application built for speed and simplicity.

Features:

  • 🔎 Intuitive Input Field: Users can enter research questions naturally.
  • ⚡ Asynchronous Communication: Smooth interaction with backend API.
  • 📘 Answer Display with Citations: Clearly formatted answers with numbered inline citations and clickable links.
  • 🧼 Minimalist UI: Optimized for clarity, accessibility, and ease of use.

Technologies:

  • Vite
  • React.js
  • HTML5 & CSS3
  • Axios / Fetch API

🖼️ Screenshots

Input Interface

Screenshot 2025-06-01 203553

Output Example

Screenshot 2025-06-01 204534 Screenshot 2025-06-01 220948


🎥 Video Demo

Watch the demo here:


🛠️ How It Works

  1. User submits a research query via the frontend.
  2. The backend sends the query to SerpAPI for web results.
  3. Relevant snippets are extracted and cleaned.
  4. OpenAI GPT generates a summary with numbered citations.
  5. Response with citations is sent back and rendered in the UI.

About

full-stack research assistant (FastAPI + OpenAI + SerpAPI) with live web search, citation generation, and content filtering to reduce hallucinations.

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published