Dooks

🚀 Inspiration As someone who loves reading and learning, I often highlighted insights that later got buried and forgotten. I craved more than just passive consumption — I wanted interaction, reflection, and continuity of thought.

What if you could talk to a book? What if every highlight triggered a deeper journey a summary, an author-style response, or even new questions to explore?

That spark led to Dooks.online an AI-powered reading companion that transforms books into dynamic, living experiences.

💡 What It Does Dooks brings books to life through AI-enhanced interactions:

Upload any book (PDF/EPUB)

Highlight passages to:

Get instant summaries and deeper insights

Chat with an AI modeled after the author’s tone

Receive semantic connections and curiosity nudges

Export markdown notes or even code with Codex

Organize highlights, ideas, and conversations in one place

See how ideas interlink across books using semantic memory

It’s more than a reader. Dooks is a thought partner.

🔧 How We Built It Bolt.new for rapid prototyping and structured execution

Netlify for frontend deployment and hosting

Supabase for database, authentication, and real-time sync

Llama for semantic search and contextual memory

ChatGPT for summarization, author-style interaction, and idea generation

OpenAI Codex for transforming highlights into structured outputs like notes, scripts, and markdown

The core loop: User highlights → Trigger AI workflows → Output insight, response, or related thought → Store and display via Supabase.

🧱 Challenges We Ran Into Maintaining author-style tone in generated responses while avoiding hallucination

Handling large book files with limited context window for accurate summarization

Creating a fast, seamless experience across multiple API calls without delay

Balancing user control (highlights, context) with AI autonomy (suggestions, prompts)

🏆 Accomplishments That We're Proud Of Built a full working MVP in a week with live demo at https://dooks.online

Seamlessly integrated ChatGPT and Codex for a multi-layered reading experience

Developed a system that makes reading more engaging, reflective, and memorable

Created a framework that can scale across genres, disciplines, and user types

📚 What We Learned Prompt design and chunking are just as critical as model choice when working with LLMs

Codex has powerful applications outside of coding, especially for structured learning

A simple frontend powered by the right backend tools (like Supabase) enables high-speed iteration

Readers are learners, and AI can enhance the relationship between text, thought, and memory

🚀 What's Next for Dooks Adding annotation layers with semantic tags and topic clusters

Supporting voice input and audio response for auditory learners

Bringing in collaborative learning: shared books, AI book clubs, group highlights

Building a personal knowledge graph that evolves as you read more

Partnering with authors and publishers to offer enhanced, AI-native editions of books

Built With

• bolt.new
• chatgpt
• codex
• llama
• netlify
• supabase