Building an AI Agent or LLM SaaS from scratch usually means spending the first 4–6 weeks wiring up the same boilerplate: authentication, multi-tenant workspaces, encrypted API key storage, a streaming canvas, and a visual workflow builder.

To skip that repetitive setup, I built NexusAI — a production-ready React 19 + TypeScript + Express + Prisma starter kit designed specifically for AI agent platforms and workflow automation tools.

✨ Key Architecture & Features

  • Visual DAG Workflow Builder: Interactive node-based flow editor with conditional branching (True / False routing), real-time step execution tracking, and run history.
  • Split-Screen AI Canvas: Live code/component generation alongside an interactive preview pane with version history (v1, v2, etc.) and file export.
  • 3-Tier AI Execution Engine: Out-of-the-box support for Google Gemini, OpenAI, and Anthropic Claude — plus an automatic Simulated Fallback Mode so every workflow runs immediately without needing external API keys during local testing.
  • Enterprise Security & Multi-Tenancy: JWT cookie authentication, RBAC workspace permissions, audit logs, and AES-256-GCM encrypted provider key storage.
  • Zero-Config Database: Ships with SQLite + Prisma seeding for instant npm run dev setup, and switches cleanly to PostgreSQL for production.

🚀 Try the Live Interactive Demo

You can explore the landing page, switch between all 3 live themes (Dark, Light, Oceanic Blue), and test the interactive workspace directly in your browser:
👉 Live Interactive Demo

📦 Get the Source Code

Available as both a lightweight Frontend UI Kit and a complete Full-Stack Starter Kit:
👉 Explore NexusAI Starter Kit on Gumroad

I'd love to hear your feedback on the architecture or any features you'd like to see added in v2!