inliver233/Ai-Novel

What it solves

Ai-Novel Lite addresses the difficulty of maintaining consistency and structure in long-form AI-assisted novel writing. Instead of treating AI as a simple text generator, it provides a structured environment to manage the complex interdependencies of a story, such as character arcs, world-building, and plot outlines, preventing the AI from losing track of the narrative over time.

How it works

The platform structures the creative process into manageable objects—projects, characters, outlines, and chapters—which are stored in a database (PostgreSQL with pgvector). It uses a combination of structured data and Retrieval-Augmented Generation (RAG) via vector search to provide the LLM with precise context and memory. Users can configure different LLM profiles and fine-tune prompts through a dedicated Prompt Studio to optimize the output for specific writing tasks.

Who it’s for

It is designed for individual authors and creative teams who want a self-hosted, professional AI writing workbench for long-term novel projects, moving beyond simple chat interfaces to a full production workflow.

Highlights

• End-to-End Workflow: Covers everything from project initiation and outlining to character management and chapter generation.
• Structured Memory: Uses vector retrieval and structured entries to maintain world-building and character consistency.
• Prompt Studio: A visual environment for debugging and managing prompt templates and task presets.
• Flexible LLM Integration: Supports multiple model configurations and parameter presets tailored to different writing tasks.
• Self-Hosted Deployment: Easy setup via Docker Compose with built-in PostgreSQL and pgvector support.