How I built an AI-first No-Code engine that actually understands your data schema

Antony Lu — Mon, 27 Apr 2026 03:56:06 +0000

Beyond Code Gen: Building Zenku, a Multi-Agent No-Code Engine

AI code generation is everywhere. You ask for a component, and it spits out 500 lines of React. But what if you need a full system? A database schema, dynamic UI views, business logic, and third-party integrations?

That's why I built Zenku. 🚀

What is Zenku?

Zenku is an open-source, AI-first No-Code engine. It doesn't just write code; it coordinates a team of specialized AI agents to build and evolve your application metadata in real-time.

antonylu0826 / zenku-v2

AI-first No-Code engine for building production-ready data apps via conversation. Features Multi-Agent architecture, dynamic UI rendering, and built-in Undo/Time-Machine. Supports SQLite, Postgres, and MSSQL.

Zenku

Build Production-Ready Data Apps via Conversation.

繁體中文 | Documentation

Zenku is an AI-first, open-source No-Code Engine designed for building enterprise-grade data applications. Instead of generating static code, Zenku uses a sophisticated Multi-Agent Architecture to dynamically evolve your database schema, UI views, and business logic in real-time through natural language.

Chat-to-App Workflow: Transform "I need a CRM" into a live system (Schema -> CRUD UI -> Dashboard) in seconds.
Specialist Agent System: Dedicated agents for Schema, UI, Logic, Query, and Testing, coordinated by a central Orchestrator.
Enterprise DB Support: Native support for SQLite, PostgreSQL, and Microsoft SQL Server (MSSQL).
Versatile View Engine: Dynamic rendering for Kanban, Gantt, Calendar, Timeline, Dashboard, and more.
Real-time Logic: A powerful Business Rules Engine for automation, validation, and third-party webhooks (e.g., n8n).
Reliability & Undo: A built-in Design Journal acting as…

View on GitHub

The Core Problem: AI Memory and Accuracy

The biggest challenge with AI builders is that they often "forget" the existing schema or make destructive changes without warning. To solve this, I implemented three core concepts in Zenku:

1. The Multi-Agent Orchestrator 🧠

Instead of one massive prompt, Zenku uses a central Orchestrator that delegates tasks to:

Schema Agent: Manages DDL and relations.
UI Agent: Renders Kanban, Gantt, and Dashboards dynamically.
Logic Agent: Handles triggers and automation rules.
Query Agent: Translates natural language to read-only SQL.

2. The "Time Machine" (Design Journal) ⏳

Every AI-driven change is recorded in a Design Journal. This allows for a robust Undo mechanism. If the AI messes up the schema, you simply say "Undo," and Zenku reverts the SQL and UI metadata to the previous state.

3. Dynamic Rendering vs. Code Gen

Zenku doesn't generate static files. It serves a dynamic JSON-defined interface. This means you can change your entire UI layout mid-conversation without waiting for a re-compile or deployment.

Technical Stack

Backend: Node.js, Express, TypeScript.
Frontend: React 19, Vite, Tailwind CSS, shadcn/ui.
Databases: Native support for SQLite, PostgreSQL, and MSSQL.
Automation: Direct integration with n8n via webhooks.

Why Open Source?

I believe the future of enterprise software is self-hostable AI. Zenku is designed to be deployed via Docker, keeping your data and your AI prompts under your own control.

I’m currently at version v1.0.0 and looking for feedback from the community. How are you using AI to automate your internal tools?

Check out the repo here: GitHub - antonylu0826/zenku-v2

I'd love to hear your thoughts in the comments!

DEV Community: Antony Lu