Skip to content

Latest commit

 

History

1 Commit

Folders and files

NameName
Last commit message
Last commit date
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Lexon AI

Transparent answers backed by verifiable evidence

MadHacks 2025 Finalist

Live Demo · Devpost · Architecture · Design Decisions


Animated Lexon AI evidence graph

Overview

Lexon AI is a source-grounded research interface that generates transparent answers, connects individual claims to supporting evidence, and visualizes the relationship between questions, answers, and sources through an interactive 3D evidence graph.

Most AI interfaces provide a response followed by a list of links. Lexon AI makes the evidence structure itself explorable. Users can inspect why a source supports a claim, compare evidence across answer blocks, and move between visual layouts designed for different research workflows.

Product Experience

Capability What it enables
Source-grounded generation Produces answers from retrieved web context rather than presenting unsupported output.
Claim-level evidence links Connects individual answer blocks directly to the sources that support them.
Interactive 3D exploration Supports rotation, zooming, inspection, filtering, and graph expansion.
Multiple layouts Provides cluster, circular, globe, baseline, and Knowledge Deck views.
Expandable research paths Extends selected sources into related concepts and research directions.
Evidence ranking Preserves retrieval relevance information throughout the answer process.

How It Works

  1. A user submits a research question.
  2. Exa retrieves relevant web sources.
  3. The answer service generates structured answer blocks from the retrieved context.
  4. The graph builder connects the question, claims, sources, and supporting concepts.
  5. The frontend renders the result as an interactive research interface.

Lexon AI system architecture

See Architecture for the request flow and Design Decisions for the graph model, layout rationale, and demo behavior.

Project Structure

Lexon-AI/
├── backend/          Retrieval, answer generation, and graph construction
├── frontend/         React and Three.js research interface
├── docs/             Architecture and design rationale
├── readme_assets/    Project visuals
├── LICENSE
└── README.md

Local Development

Requirements

  • Node.js 18 or newer
  • npm
  • Exa API credentials
  • OpenAI API credentials

Installation

git clone https://github.com/Tejas7007/Lexon-AI.git
cd Lexon-AI
npm run install:all

Create backend/.env from the provided example and add your credentials. Create frontend/.env from frontend/.env.example and configure the backend URL.

Start the backend:

cd backend
npm run dev

Start the frontend in a second terminal:

cd frontend
npm run dev

Run the deterministic backend tests and both production builds:

cd backend
npm test
cd ..
npm run build

Team

Lexon AI was developed at MadHacks 2025 by:

  • Tejas Dahiya
  • Quinn Hasse
  • Franco Pastor Cueto
  • Ares Stavropoulos

License

Released under the MIT License.

About

Lexon AI (MadHacks 2025 finalist) is a source-grounded research interface that generates transparent answers, connects every claim to verifiable evidence, and visualizes the relationships between questions, answers, and sources through an interactive 3D evidence graph.

Topics

Resources

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages