Inside Risentta Write: Engineering a Real-Time CRDT & Multi-Model AI Collaborative Editor

As Founder & CTO at Risentta (www.risentta.com), I architected a unified Next.js 16 (App Router) monorepo comprising 13 MongoDB domain models, 50+ Radix UI components, and 9+ API domains to power our public brand platform, internal multi-division RBAC admin dashboard, customer SaaS portal, and flagship AI collaborative workspace.

1. Conflict-Free Real-Time Collaboration with Y.js & WebSockets

At the heart of Risentta Write is a multi-user rich-text editor built on Tiptap and synchronized via Y.js CRDTs (Conflict-free Replicated Data Types) over WebSockets:

• Live Multi-User Cursor Presence: Real-time cursor positions and selection highlights per collaborator.
• 20+ Formatting Capabilities & Academic Tools: Integrated citation manager, document version history, structured templates, in-document team chat, and a synchronized Virtual Speaker + YouTube queue for collaborative study/research sessions.

2. Dual-Engine AI Orchestration (Google Gemini + Claude via OpenRouter)

To empower researchers and writers, I built a unified AI orchestration layer connecting Google Gemini API and Anthropic Claude (via OpenRouter):

• Contextual Inline AI Actions: Paraphrase, summarize, and expand selected text blocks directly inside the editor canvas.
• Single-Prompt Full Document Generation: Auto-draft complete, structured academic and analytical documents from a single research prompt.
• Airisenta AI Chat: A standalone conversational research assistant with persistent chat history and subscription-linked token quota gates (Free, Daily, Weekly, Biweekly).