Gobook

Inspiration

Every parent has heard the words, "Read me a story!" We wanted to make that experience limitless. Instead of choosing from a fixed library of books, we imagined a platform where any idea could become a personalized children's storybook. Large language models have become excellent storytellers, but turning a simple prompt into a beautifully illustrated, narrated, and printable book still required stitching together multiple AI services into one seamless experience. GoBook was built to bridge that gap.

What it does

GoBook transforms a single prompt into a complete children's storybook. The application first generates a structured story tailored to a child's age range, then creates a consistent cover and page illustrations, generates narration for every page, and finally composes everything into downloadable PDF and ePub formats, as well as an interactive browser-based flipbook. Throughout the generation process, users receive live progress updates so they can watch each stage complete in real time.

How we built it

GoBook is powered by Genblaze for pipeline orchestration and Backblaze B2 for persistent object storage. Genblaze coordinates every stage of the AI workflow—from story planning, illustration generation, and narration to final book composition. Each step runs independently, streams progress updates to the frontend through Server-Sent Events (SSE), and integrates multiple AI providers behind a single pipeline interface. We used Groq for story generation, Cloudflare Workers AI (FLUX/Stable Diffusion XL) for illustrations, and ElevenLabs for text-to-speech narration. Every generated asset—including story plans, images, narration audio, metadata, and finished books—is stored in Backblaze B2 using its S3-compatible API. The application retrieves these assets with secure presigned URLs to serve previews, downloads, and compose the final book.

Challenges we ran into

Building a reliable multi-modal AI pipeline was our biggest challenge. Each AI provider has different response times, APIs, and failure modes, so coordinating text generation, image generation, narration, storage, and document composition required careful orchestration. Maintaining a consistent illustration style across multiple pages was another challenge, as was generating long-running content while keeping users informed through live progress updates. We also optimized storage and asset delivery using Backblaze B2 to keep costs low while ensuring every generated book could be retrieved and shared later.

Accomplishments that we're proud of

We're proud that GoBook can generate an entire illustrated children's book—from a single prompt—with consistent artwork, narration, and downloadable formats. We're also proud of the architecture behind it. By combining Genblaze's pipeline orchestration with Backblaze B2's scalable storage, we built a modular system where AI providers can be swapped or extended with minimal changes. The real-time streaming interface also gives users visibility into every generation stage instead of waiting behind a loading spinner.

What we learned

This project taught us that building AI applications is less about calling models and more about orchestrating reliable workflows. Breaking the generation process into independent pipeline stages made the system easier to scale, debug, and extend. We also learned the value of separating compute from storage. Using Backblaze B2 as the central storage layer simplified asset management and made it easy to compose final books from previously generated content.

What's next for Gobook

We plan to add support for additional AI providers, multilingual story generation, customizable narration voices, and collaborative story creation. We're also working on reducing generation times through parallel pipeline execution and adding user accounts, saved libraries, and story sharing. Long term, we want GoBook to become a platform where parents, educators, and children can instantly create personalized books that can be read online, downloaded, printed, or narrated in multiple languages.

Built With

• ai
• b2
• backblaze
• cloudflare
• css
• elevenlabs
• fastapi
• flux.1
• fpdf2
• genblaze
• groq
• llama
• neon
• next.js
• python
• react
• react-pageflip
• render
• schnell
• tailwind
• typescript
• vercel
• workers