Zoonk - AI Learning Platform Setup

TL;DR

TL;DR: Zoonk is an open-source AI-powered learning platform that generates structured, interactive courses from any topic. Built with Next.js, TypeScript, Prisma, and PostgreSQL, it runs as a pnpm monorepo and is deployable on Vercel.

Source and Accuracy Notes

• Project page: zoonk.com
• Source repository: github.com/zoonk/zoonk
• License: MIT (verified via GitHub API license.spdx_id)
• HN launch thread (original project as Uneebee): news.ycombinator.com/item?id=38316936
• Source last checked: 2026-06-17 (commit on main branch, no versioned releases yet)

What Is Zoonk?

Zoonk is an AI-powered learning platform that turns any topic into clear, structured lessons. It started as Uneebee (an open-source Duolingo alternative for creating interactive courses) and has been rewritten from the ground up with a new focus: using LLMs to generate personalized learning paths.

From the project’s README:

Turn any topic into clear, structured lessons.

The platform offers:

• Personalized lessons tailored to your goals, level, and interests
• 70 languages with pronunciation guidance
• Guided courses with progress tracking
• Exam preparation with simulations and quizzes
• Interactive exercises instead of passive video watching

Tech Stack

Zoonk is a TypeScript monorepo managed with pnpm workspaces and Turborepo:

| Component | Technology | |-----------|-----------| | Runtime | Node.js v24 | | Package manager | pnpm v11 | | Framework | Next.js (App Router) | | ORM | Prisma | | Database | PostgreSQL v18 | | UI | shadcn/ui + Tailwind CSS | | AI | Vercel AI SDK | | Auth | Better Auth | | Background jobs | Workflow (useworkflow.dev) | | Linting | oxlint + oxfmt | | Testing | Vitest + Playwright | | Hosting | Vercel (OSS program) |

Prerequisites

Before setting up Zoonk locally, ensure you have:

• Node.js v24 (use mise or nvm to manage versions)
• pnpm v11 (corepack enable or npm install -g pnpm@11)
• PostgreSQL v18 (see setup below)

Setup Workflow

Step 1: Clone and install dependencies

git clone https://github.com/zoonk/zoonk.git
cd zoonk
pnpm install

Step 2: Set up PostgreSQL

On macOS with Homebrew:

brew install postgresql@18
brew services start postgresql@18

# Add to PATH if needed
export PATH="$(brew --prefix postgresql@18)/bin:$PATH"

# Create the database
createdb zoonk

# Create test database (optional)
createdb zoonk_test

The test setup expects a postgres user with password postgres. Create it:

psql postgres

CREATE USER postgres WITH PASSWORD 'postgres' SUPERUSER;
\q

Step 3: Configure environment variables

cp .env.example .env

Edit .env with your database connection string and API keys. Check packages/db/README.md for database-specific variables and apps/main/README.md for app-specific setup.

Step 4: Generate Prisma client and run migrations

pnpm db:generate

Step 5: Start the development server

pnpm dev

The app runs via Turborepo, which starts all workspace apps in parallel.

Step 6: Run tests

pnpm test          # Unit tests (Vitest)
pnpm e2e           # End-to-end tests (Playwright)
pnpm typecheck     # TypeScript checking
pnpm lint          # Linting (oxlint)

Deeper Analysis

From Uneebee to Zoonk

The original Uneebee project (1,300+ stars) was built with Elixir/Phoenix and focused on course creation. The README now carries a caution notice:

Uneebee is no longer being maintained. I’m building a new platform, Zoonk, with a different focus.

The rewrite swaps Elixir for the TypeScript/Next.js ecosystem, trading Phoenix LiveView’s real-time capabilities for the broader Next.js/Vercel deployment story. The new architecture uses:

• Vercel AI SDK for LLM integration (course generation, personalized content)
• Better Auth for authentication (replacing Phoenix’s built-in auth)
• Workflow for durable background jobs (AI generation tasks)
• Prisma as the ORM layer over PostgreSQL

Monorepo Structure

zoonk/
├── apps/
│   ├── main/          # Next.js app (zoonk.com)
│   └── evals/         # AI evaluation test suite
├── packages/
│   ├── auth/          # Better Auth config
│   ├── db/            # Prisma schemas + client
│   └── ...            # Shared packages
├── turbo.json         # Turborepo config
└── package.json       # pnpm workspace root

AI-First Approach

Unlike traditional LMS platforms (Moodle, Canvas), Zoonk generates content on-demand. You describe a learning goal, and the AI produces:

1. A structured course outline
2. Individual lessons with explanations
3. Quizzes and exercises
4. Progress tracking

This is closer to Khanmigo (Khan Academy’s AI tutor) than to course marketplaces like Udemy.

Practical Evaluation Checklist

Before self-hosting Zoonk for production use:

• [ ] Node.js v24 requirement — this is cutting-edge; most hosting providers default to v20/v22. Verify your deployment target supports v24.
• [ ] PostgreSQL v18 — similarly bleeding-edge. Neon, Supabase, and RDS may not yet offer v18.
• [ ] AI API costs — Zoonk uses the Vercel AI SDK, which supports multiple providers (OpenAI, Anthropic, etc.). Estimate per-user generation costs.
• [ ] No versioned releases — the project tracks main with no tags. Expect breaking changes without migration guides.
• [ ] Vercel dependency — the project is in Vercel’s OSS program and likely optimized for their platform. Self-hosting on AWS/GCP may require additional config.
• [ ] Background jobs — Workflow (useworkflow.dev) is a newer durable execution framework. Evaluate its maturity for production workloads.

Security Notes

• Environment variables — never commit .env files. The repo uses .env.example as a template.
• Database credentials — the default test setup uses postgres/postgres. Change this in production.
• Auth — Better Auth handles session management. Review its security documentation before deploying.
• AI-generated content — LLM outputs can contain hallucinations or inappropriate content. Implement content filtering for public-facing deployments.
• Dependency auditing — run pnpm audit regularly. The monorepo has many dependencies across workspaces.

FAQ

Q: Is Zoonk free to use? A: The source code is MIT-licensed and free to self-host. The hosted version at zoonk.com offers a free tier (no credit card required) with paid plans for additional features.

Q: Can I use Zoonk without AI? A: No. Zoonk’s core value proposition is AI-generated content. The platform requires LLM API access (OpenAI, Anthropic, etc.) to generate lessons and courses.

Q: What happened to Uneebee? A: Uneebee (the original Elixir/Phoenix project) is no longer maintained. The developer rewrote it as Zoonk with a new tech stack (TypeScript/Next.js) and a shifted focus from course creation to AI-generated learning paths. The Uneebee repo remains on GitHub for archival purposes.

Q: Can I deploy Zoonk on my own server? A: Yes, but it requires Node.js v24 and PostgreSQL v18, which are very recent versions. The project includes a Dockerfile for Fly.io deployment, but Vercel is the recommended platform.

Q: Does Zoonk support multiple languages? A: The platform supports 70 languages for learning (as course content). The UI is internationalized using Next.js i18n with translation files in apps/main/messages/.

Q: How does Zoonk compare to Duolingo? A: Duolingo focuses on language learning with gamification. Zoonk is a general-purpose learning platform that can generate courses on any topic, not just languages. The original Uneebee project was explicitly positioned as a “Duolingo for X” creator tool.

Conclusion

Zoonk represents an interesting evolution in open-source education tools. By combining AI generation with structured learning paths, it moves beyond traditional LMS platforms that require manual course creation. The TypeScript/Next.js stack makes it accessible to a large developer audience, and the Vercel OSS program membership signals production readiness.

However, the bleeding-edge requirements (Node.js v24, PostgreSQL v18) and lack of versioned releases mean this is best suited for early adopters and contributors, not production deployments mission-critical learning infrastructure. If you’re building an AI-powered education product or want to experiment with LLM-generated courses, Zoonk’s MIT-licensed codebase is worth exploring.

Star the repo on GitHub at github.com/zoonk/zoonk and try the hosted version at zoonk.com.