Building Novelist: An AI-Native Novel Writing Platform on AWS
Create AI novels with illustrations (or) Bring your own story and get it illustrated using AI
Building Novelist: An AI-Native Novel Writing Platform on AWS
Over the past few days, I challenged myself to answer a simple question:
What if writing a complete novel felt less like prompting an AI... and more like collaborating with an intelligent writing partner?
That question became Novelist.
Rather than building another "chat with an LLM" application, I wanted to design a system that treats AI as one component of a larger architecture—one that's scalable, secure, and production-ready.
The idea
Novelist is an AI-powered platform that helps authors generate stories, characters, scenes, world-building, illustrations, and complete manuscripts through a structured workflow instead of a single chat window.
The focus wasn't just on generating text.
It was about building the infrastructure that could support an entire creative pipeline.
Tech Stack
Frontend
- Next.js 16 App Router
- Vercel Edge Runtime
- Better Auth
- Server Components
- Edge Functions
Backend
- Model Context Protocol (MCP) Server
- Amazon Bedrock with fallback to 3rd party model when rate limits apply
- Amazon Aurora Serverless v2 (PostgreSQL)
- Amazon S3
What I Learned
The biggest realization wasn't about prompting.
It was about orchestration.
As applications begin using multiple foundation models, prompt engineering becomes only one piece of the puzzle.
Managing context.
Managing workflows.
Managing state.
Managing assets.
Those become the real engineering challenges.
That's exactly where protocols like MCP start becoming incredibly valuable.
What's Next
There are several features I'd love to explore next:
- Character memory
- Story consistency validation
- AI-powered illustration refinement
- EPUB and PDF publishing
- Collaborative author workspaces
Building Novelist reminded me that great AI applications aren't just about choosing the best model.
They're about designing systems where models, data, orchestration and user experience work together seamlessly.
I'm looking forward to continuing to iterate on this project and would love feedback from fellow builders.
If you're working with Amazon Bedrock, MCP, or AI-native application architectures, I'd love to connect and exchange ideas.
Enjoyed reading this content? Let the author know!
Your likes, comments, shares, and saves help creators reach more builders.
Loading recommendations
Loading article