Back to Home
Every backend developer knows the drill. You spin up a new Postgres database, add a handful of tables, and then face the same tedious ritual: scaffold a Next.js project, wire up an ORM, hand-build forms and data tables for every entity, set up a GitHub repo, configure CI/CD, and finally get something deployed so a teammate can actually poke at the data.
That's the itch instant-crud-panel (InstantPanel) scratches. It's an AI-powered "application factory" that takes a live PostgreSQL connection and turns it into a deployed, full-stack CRUD app, without a human writing a single React component along the way.
InstantPanel's core idea is a straight line from database to deployed app: PostgreSQL connection to schema introspection to table selection to GitHub repository creation to AI code generation to PR merge to Netlify deployment to a live application.
You point it at a Postgres database, it introspects the schema, and you pick which tables you want an interface for. From there it provisions a GitHub repository, generates the actual Next.js application code with AI, merges the resulting pull request, and deploys the result to Netlify, complete with environment variables already wired up. What starts as a connection string ends as a URL you can click.
The README is upfront about the gap it's trying to close. Tools like Retool solve the "build an internal CRUD tool fast" problem well, but they come with per-seat subscription pricing and no self-hosted source code, you're renting the tool, not owning it. InstantPanel's answer is to generate a real, ordinary Next.js codebase that lives in your own GitHub account and deploys to your own Netlify account, so what you end up with is a normal app you can read, modify, and host yourself rather than a black box behind a paywall.
The stack is a fairly modern, opinionated one: Next.js 16 with the App Router, Clerk for authentication, Prisma as the ORM, and Google's Jules API doing the actual code generation. GitHub and Netlify APIs handle repo provisioning and hosting respectively.
What's more interesting than the stack is the multi-agent system driving the generation step. InstantPanel splits the work across a small cast of autonomous agents: a Scout that does market research and proposes features, an Architect (running on Gemini 3.1 Pro) that turns proposals into specs and GitHub issues, a Builder (Jules) that writes the actual Next.js code and opens pull requests, and a Sentinel that reviews the generated code, checks it for security issues, and handles deployment. It's less "click a button and an LLM writes some code" and more a small assembly line of specialized agents, each responsible for one stage of the process.
There's even a house design system baked into the generated apps, cheekily named "The Kinetic Monolith", it favors elevation through background color shifts over borders, uses Material Symbols Outlined icons, and leans on Manrope for type. It's a nice touch that the generated apps aren't generic Bootstrap-looking scaffolds but have an actual point of view.
Getting a local instance running is standard Next.js fare: clone the repo, run npm install, copy .env.example to .env.local and fill in Clerk credentials, a DATABASE_URL, a GitHub personal access token with repo scope, a Jules API key, and Netlify credentials. Run npx prisma migrate dev to set up the database, then npm run dev, sign in through Clerk, and head to /dashboard to start connecting databases.
The project is honest about being an MVP, and the README calls out two limitations worth knowing before you point it at anything important. Generated applications and their deployments currently live under the platform provider's own GitHub and Netlify accounts rather than the end user's, and the generated CRUD apps don't ship with authentication out of the box, which means a generated app's database access could be left exposed if you don't lock it down yourself. Both are the kind of thing you'd expect from an early-stage project, and the roadmap implied by the multi-agent Sentinel role, security verification is already part of its job, suggests they're on the radar.
There's a broader trend here that InstantPanel is a good example of: using AI not to write code snippets on demand, but to run an entire scaffold-to-deploy pipeline autonomously, with different agents responsible for different concerns the way a small team would divide labor. Internal tools and admin panels are a great proving ground for this, the requirements are well-understood, the patterns are repetitive, and the cost of a wrong turn is low. If InstantPanel (or something like it) matures past its MVP limitations, "give me a CRUD app for this table" could become as routine as running a database migration.
Repo Here -> https://github.com/arif-js/instant-crud-panel