Introducing GoLiveKit: an AI-driven Next.js SaaS starter kit
Over the last several years, I’ve built SaaS products across different stacks, deployment models, and team setups. During a sabbatical period, I finally had enough focused time to consolidate what actually worked into one reusable starter kit.
GoLiveKit is the result: a production-ready foundation for building and self-hosting modern SaaS apps with a stack that feels current, practical, and optimized for fast iteration.
One codebase, practical defaults, AI-aware context, and a deployment path that does not require a large platform bill.
Why I built it
My first SaaS projects were built with PHP and Laravel. Later I moved into Node.js with Express, then worked with Vue and React on the frontend. After that came a NestJS + Next.js setup, which worked well, but maintaining separate backend and frontend apps, deployments, and infrastructure added a lot of overhead.
With modern full-stack Next.js, especially in the latest versions, that complexity can be reduced significantly. You can keep the product in one codebase, move faster, and still build something structured enough for real production use.
That shift is what led me to package everything I’d been repeating from project to project into GoLiveKit.
The 3 core principles behind GoLiveKit
| Principle | What it means in practice |
|---|---|
| 🤖 AI readiness | AI agents get project-specific context, so they can make better changes with fewer roundtrips. |
| 🐳 Easy self-hosting | Deploy to a VPS with Docker and GitHub Actions instead of stitching together custom infrastructure each time. |
| 💸 Cost optimization | Keep hosting simple, predictable, and as vendor-independent as possible. |
1. 🤖 AI readiness
Every part of the kit is documented with AI context through dedicated SKILL.md files and supporting instructions. The goal is simple: help tools like Claude, Copilot, and Cursor understand the codebase faster, reduce roundtrips, and generate more accurate changes.
This makes the starter kit especially useful if AI is already part of your development workflow.
2. 🐳 Easy self-hosting on a VPS
I wanted deployment to be straightforward and affordable, without turning every launch into a DevOps project.
GoLiveKit is prepared for self-hosting on a VPS through Docker and GitHub Actions. You can deploy to providers like DigitalOcean, Hetzner, or AWS EC2 using a workflow that is much simpler than manually configuring servers from scratch.
The Docker-based setup includes the core services you usually need:
- Traefik for reverse proxying and automatic SSL
- PostgreSQL as the primary database
- Dozzle for live log monitoring
Dozzle and Adminer are protected behind Traefik basic auth by default, and the GitHub Actions pipeline handles linting, building, and deployment.
3. 💸 Cost optimization
I strongly prefer predictable infrastructure costs and minimal vendor lock-in.
In many cases, all you need to run a project built on GoLiveKit is:
- a domain name, usually around $10/year
- a small VPS, often around 12/month
For file storage, the kit supports AWS S3 and Cloudflare R2. That is one of the few places where I intentionally rely on external providers because it simplifies the architecture and both options have generous free tiers.
The stack
I built GoLiveKit around a stack I genuinely like using today, with an emphasis on developer experience, maintainability, and shipping speed.
| Layer | Tools |
|---|---|
| AI | SKILL.md, Claude, Copilot, Cursor |
| Backend | Next.js, Prisma, oRPC, TypeScript, Zod, next-intl |
| UI | Tailwind CSS, shadcn/ui, Plate.js |
| Tooling | GitHub Actions, Docker, Docker Compose, pnpm, Biome |
AI
- AI-aware project context through
SKILL.mdfiles for every major part of the codebase - Better support for AI-assisted coding in Claude, Copilot, and Cursor
Backend
- Next.js for full-stack application development
- Prisma ORM for database access
- oRPC for OpenAPI, RPC, and server action workflows
- TypeScript and Zod for type-safe app logic and validation
- next-intl for localization support
UI
- Tailwind CSS
- shadcn/ui
- Plate.js for rich text and markdown-style editing experiences
CI/CD and tooling
- GitHub Actions for linting and VPS deployment
- A prepared Dockerfile using Next.js standalone output
- Docker Compose with PostgreSQL and live log monitoring services
- pnpm for package management
- Biome for formatting and linting
What’s included out of the box
Prebuilt integrations
- Payments: Stripe support for subscriptions, one-time purchases, and credit-based flows. Docs
- Storage: AWS S3 and Cloudflare R2 integration. Docs
- Emails: Resend, Plunk, and Nodemailer support. Docs
- Analytics: Google Analytics integration. Docs
- Notifications: Telegram alerts for app events. Docs
- Authentication: Passwordless magic links and Google login. Docs
Product features
- Admin panel for managing users, blog posts, and pages. Docs
- Dashboard foundation for authenticated product areas
- Payments flow with subscriptions, one-time payments, and credits usage. Docs
- Blog with SEO-friendly publishing support. Docs
- Pages for content like Terms of Service and Privacy Policy. Docs
- Built-in documentation powered by Fumadocs. Docs
- SEO setup with metadata, descriptions, and generated Open Graph assets. Docs
- Cron jobs and log monitoring for background tasks and operational visibility. Docs
Getting started
- Local development: Start the app locally with the full development environment. Docs
- Environment variables: Configure the project without guesswork. Docs
Building with it
- AI skills: Learn how to use the project’s AI context effectively. Docs
- AI tasks: Plan implementation scope more efficiently. Docs
- APIs: Build structured APIs with the included patterns. Docs
- UI components: Work with reusable UI primitives, filters, tables, forms, and editor tooling. Docs
Deploying to a VPS
- Docker VPS deployment: Launch on DigitalOcean, Hetzner, or AWS EC2. Docs
- Terraform: Provision infrastructure faster with reusable templates. Docs
- Log monitoring: Inspect runtime logs with the included monitoring setup. Docs
Understanding the codebase
- Project structure: Learn how the application is organized. Docs
- Formatting and linting: Biome-based formatting and linting setup. Docs
- VS Code setup: Recommended editor configuration and extensions. Docs
- Updating the codebase: Keep your fork aligned with new starter kit improvements. Docs
What’s next
This is the initial public release, and I plan to keep improving it.
The next areas of focus are:
- Better tutorials across docs and video
- A smoother path from first install to first production launch
- Continued iteration based on user feedback
If you’re building a SaaS product and want a practical, self-hostable, AI-friendly starting point, GoLiveKit is built for exactly that.