Supabase & Cursor with vibe coding no code background. How to build easier?
Hi, I am a 46 year old Real Estate tech guy, but I could not code until AI. Started with Bubble.io in October 2024, Replit in November, Cursor December. I have been coding with AI daily since December (founders life 😬 ). One of my main hurdles still is Cursor and Supabase. It must be a understanding, communication and gaps in knowledge on my side. I am sometimes struggling to keep the right context and make the LLM's create the right migration files.
Also, I now have a national DB structure. But we are going global. This means copies, but also exceptions etc. I would love to spar about options there. And humbly I submit my lack of IT knowledge and even fully understanding POSTGRES or SQL for that matter. Yet I am very effective and creative with my team of AI coding assistants in an out of Cursor.
I am to be a very big user with our platform aiming to save images for every house on the planet. But since every time there is supabase involved it takes a lot of time, and I am always happy that everything seems to work. That could improve :). Also, perhaps my experience as interested user but with a knowledge gap can give some helpful insights. (I feel like I use Supabase like a caveman)
4 Replies
For isntance. This is a plan Sonic (Grok 4 coding) in Cursor created in 6 steps to add businessess options. But the first go. I start off with: ERROR: 23514: check constraint "user_profiles_role_check" of relation "user_profiles" is violated by some row off to a bad start, the first gave this error 🙂 .... And I am used to errors, AI's used to hallucinate and smaller context so it is already way way better. And switching models helps. And yet here I am, switched models. it is some stupid thing with my current user profiles. I will overcome this. But I would love to learn a way where the context is 100% aligned and the AI can easily adapt and change. (and no I wont let Claude anywhere near my database, drop all hahaha).
Thanks for sharing your journey! Looks like you have a great adventure ahead!
In order to understand better your current stack and see what resources could be helpful, some questions:
- What tech stack are you using? NextJS/React + Vite?
- Are you using Supabase MCP?
- Have you setup Supabase recommended rules for cursor? You can find some good ones here
AI Prompts | Supabase Docs
Prompts for working with Supabase using AI-powered IDE tools
These tools/resources that Supabase provides help a lot on the journey, they won't make the AI building errors-free, there is still a lot of issues around hallucinations with models
Summary: HandyHouseHelp.ai - AI-Powered Home Management Platform
Our Stack & Architecture
We're a full-stack AI-powered home management platform built for Dutch homeowners and real estate professionals. Our tech stack combines modern web technologies with advanced AI capabilities:
Frontend: React 18 + TypeScript + Tailwind CSS + Vite
Backend: Node.js + Express.js + Supabase (primary database & auth)
AI/ML: Google Generative AI + custom AI agents
Deployment: Vercel + Supabase Edge Functions
Mobile: Progressive Web App with native app experience
Core AI Features
1. Intelligent Chat System 🤖
Context-Aware AI Assistant: Our AI understands Dutch home maintenance context
Multi-Modal Input: Processes text + images for repair assessments
Real-Time Collaboration: Homeowners and real estate agents chat with AI simultaneously
Learning System: AI improves recommendations based on user interactions
2. Digital Twin Technology 🏠
Property Digital Twins: Complete virtual representation of Dutch homes
BAG Integration: Automatic data enrichment from Dutch government databases
Timeline Intelligence: Predicts maintenance needs based on property age and location
Visual Inspection AI: Analyzes photos to detect maintenance issues
3. Smart Scheduling & Planning 📅
AI-Powered Scheduling: Automatically schedules maintenance based on:
Weather conditions
Seasonal requirements
Property age and location
User availability
Conflict Resolution: Intelligently resolves scheduling conflicts
Predictive Maintenance: Proactively suggests work before issues become critical
4. Advanced Document Processing 📄
Brochure AI: Processes PDF brochures and extracts structured property data
Vision AI Pipeline:
PDF to image conversion
Object detection for property features
Text extraction and categorization
Data enrichment and validation
Real Estate Intelligence: Automated property analysis for real estate agents
short overview. will look into the tips you gave.