VerifyAI is an AI-powered metasearch app that finds relevant web pages, news, and images via Firecrawl, then generates fast, source-grounded summaries and answers using Groq-hosted LLMs. It aims to deliver concise responses with inline citations and a clean, minimal UI.
Key ideas:
-
Fetch multi-source evidence first, answer second.
-
Always cite sources, show result cards, and keep a full reference trail.
-
Optimize for latency using Groq’s low-latency inference.
-
Web, news, and image search via Firecrawl
-
Source expansion: scrape/crawl pages to get LLM-ready markdown
-
Answer synthesis using Groq LLMs with explicit grounding in fetched sources
-
Inline citations and reference list
-
Result cards with titles, snippets, and favicon/thumbnails
-
Follow-up questions and conversational context
-
Safety filters and prompt-guardrails
-
Caching for deduping identical queries
Language: TypeScript for full-stack type safety
Runtime: Node.js (server) and modern browser (client)
Frameworks: Next.js or Vite + React (SPA/SSR choice up to implementation)
Styling: CSS modules or Tailwind CSS
Data fetching: Fetch/axios with retries and timeouts
API:
-
Firecrawl API for search, crawl, scrape, images
-
Groq API for LLM summarization, answer synthesis, and citation formatting
Build Tooling: ESLint, Prettier, pnpm/npm
Install my-project with npm
# 1. Clone the repo
git clone https://github.com/harshaygadekar/VerifyAI.git
cd VerifyAI
# 2. Copy env example
cp .env.example .env
# 3. Install dependencies
# Using pnpm or npm
pnpm install
# or
npm install
# 4. Run development server
pnpm dev
# or
npm run dev
# 5. Build for production (if needed)
pnpm build
# or
npm run build
# 6. Start production server
pnpm start
# or
npm run start