Ai Facts

AI Facts

**Note:** LLMs can sometimes provide incorrect or outdated information. Always verify critical information through trusted sources.

[](https://vercel.com/new/clone?repository-url=https%3A%2F%2Fgithub.com%2Fvercel-labs%2Fai-facts&env=DEEPGRAM_API_KEY&envDescription=Learn%20more%20about%20how%20to%20get%20the%20API%20Keys%20for%20the%20application&envLink=https%3A%2F%2Fgithub.com%2Fvercel-labs%2Fai-facts%2Fblob%2Fmain%2F.env.example&demo-title=AI%20Facts&demo-description=Real-time%20fact%20checking%20using%20audio%20transcription%20and%20AI&demo-url=https%3A%2F%2Fai-facts.vercel.app)

This project is a Next.js application that performs real-time fact checking on spoken statements. It uses Deepgram for audio transcription and leverages OpenAI and Perplexity (via the Vercel AI Gateway) to verify the accuracy of claims.

Features

• Real-time Audio Transcription: Captures and transcribes spoken audio using Deepgram's API
• AI Fact Checking: Uses both OpenAI and Perplexity to cross-reference and verify statements
• Live Results: Shows fact-checking results in real-time as statements are processed
• Explanation of Validity: Provides detailed explanations for why statements are considered true or false

Technology Stack

• Next.js for the frontend and API routes
• AI SDK for interacting with LLMs
• Vercel AI Gateway for unified access to OpenAI and Perplexity
• Deepgram for audio transcription
• ShadcnUI for UI components
• Tailwind CSS for styling

How It Works

1. Speak into microphone
2. Deepgram processes the audio stream in real-time and returns transcribed text
3. The transcribed text is analyzed for distinct statements ('?!.')
4. Each statement is routed through the AI Gateway to OpenAI and Perplexity for fact checking
5. The verificat