TLDR - Document Summarizer

Inspiration

Reading long documents takes time and sustained focus. I wanted to create a tool that distills complex documents into clear summaries you can listen to while multitasking, commuting, or whenever reading isn't convenient.

What it does

TLDR transforms PDFs, text files, and markdown into concise summaries with natural AI voice narration. Users can customize summary length, add instructions like "explain like I'm 5," and download summaries for offline use.

How we built it

Built with Next.js 14, TypeScript, and Tailwind CSS. Google Gemini 2.5 Flash handles document understanding and summarization, while ElevenLabs provides natural text-to-speech. I implemented browser TTS fallback for when API credits run low.

Challenges we ran into

• Managing ElevenLabs free tier credits during testing
• Converting audio streams to browser-playable format
• Building solo in 2.5 days with power outages required aggressive prioritization
• Learning to leverage Gemini's native PDF support instead of client-side parsing

Accomplishments that we're proud of

• Fully functional AI app built solo in under 3 days
• Clean, maintainable TypeScript codebase

What we learned

Gemini's native document understanding is incredibly powerful.

What's next for TLDR - Document Summarizer

• User authentication and summary history (MongoDB + BetterAuth)
• Multiple voice options and languages
• Multi-document synthesis and comparative analysis
• Conversation mode: two AI voices discussing documents podcast-style
• Mobile app for better offline support

Built With

• elevenlabs
• gemini
• nextjs
• tailwindcss
• typescript
• webspeechapi