Loading...
Loading...
Every TypeScript developer building AI hits the same wall: the RAG ecosystem is Python-first.
Not anymore. Full production RAG in your stack — streaming, voice, auth, background jobs. Zero API costs to start.
➜ ~ git clone https://github.com/rejisterjack/rag-starter-kit.git
➜ rag-starter-kit cp .env.example .env && bun dev
[+] Installing dependencies...
✔ Connecting to Prisma Postgres...
✔ Connecting to Upstash Redis...
✔ Inngest dev server started...
🚀 Ready at http://localhost:7392
Clone, configure two free API keys, and deploy. No Python. No credit card. No tutorial hopping.
Click "Run" to see the setup animation
Built for developers who want to ship. Not for those looking for a no-code tool.
Building a product in Next.js
You need RAG, but you don't want to learn Python infrastructure or stitch together 5 different tutorials. You want a codebase that feels native to your stack.
Breaking into AI engineering
You're tired of 'hello world' tutorials. You want to see how real streaming, vector databases, and background jobs work together in a real app, without spending a dime on API credits.
Delivering client AI projects
Clients want chatbots trained on their docs. Don't build from scratch. Clone this, customize the UI, upload their docs, and deploy by Monday. Charge for value, not boilerplate.
LangChain is great — if you work in Python. If you build your product in TypeScript, you shouldn't need a separate Python service just to add AI.
| Feature | RAG Starter Kit | LangChain (Python) | LlamaIndex | From Scratch |
|---|---|---|---|---|
| Language | TypeScript | Python | Python | Depends |
| API cost to start | $0 | Varies | Varies | Depends |
| Auth included | Yes | No | No | Build it |
| Background jobs | Inngest | No | No | Build it |
| Streaming UI | SSE | Varies | Varies | Build it |
| One-click deploy | Vercel | No | No | Build it |
| Production-ready | Yes | Partial | Partial | 3 weeks |
From agency deliverables to internal tools, here is how developers are using the starter kit today.
Client wants an AI chatbot trained on their documentation. Clone the repo, upload their docs, configure two free API keys, and deploy to Vercel. Done by Monday.
Index your SaaS product's entire documentation site and help articles. Watch repetitive L1 support questions get answered automatically in real-time.
Years of scattered Confluence pages and Google Drive documents, finally indexed and searchable through a single, secure chat interface for your entire team.
A real production codebase, not a tutorial. See exactly how chunking, embedding, pgvector, and streaming SSE work together without paying for API credits.
Production infrastructure already wired in. Auth, storage, queues, analytics — it's the project you'd spend three weeks building.
Context-aware responses using LangChain.js and pgvector. Hybrid search combines vector similarity with keyword matching for unprecedented accuracy.
Upload PDFs, Word docs, Markdown, and text files. Background processing via Inngest handles large documents without blocking the UI.
Lightning-fast token streaming via Server-Sent Events. Responses appear in real-time with sub-second first token latency.
Collaborative workspaces with typing indicators, presence tracking, and role-based access. Built for teams from day one.
Speech-to-text via Web Speech API + Whisper. Text-to-speech synthesis. Wake word detection ("Hey RAG") for hands-free operation.
NextAuth.js v5 with OAuth (GitHub, Google), audit logging, rate limiting via Upstash Redis, and row-level database isolation.
No Python, no context switching. Next.js 15 App Router, React 19, Tailwind CSS 4, shadcn/ui — all strictly typed end-to-end.
Install as a native app on any device. Offline support, push notifications, and responsive design across all breakpoints.
Plausible analytics, Pino structured logging, OpenTelemetry traces. Know exactly what your AI is doing.
Save any web page to your knowledge base with one click. Ask questions about selected text, summarize pages, and search your docs — all from the browser.
Connect GitHub repos, Google Drive, Notion pages, Slack channels, and YouTube videos. Import knowledge from the tools your team already uses.
Teams shipping production RAG systems with RAG Starter Kit
“We went from zero to a production RAG chatbot in a weekend. The TypeScript stack meant our whole team could contribute immediately.”
Sarah K.
Engineering Lead, Series B Startup
“Finally a RAG kit that doesn't require Python. The pgvector integration is rock solid and the multi-provider AI support saved us weeks.”
Marcus T.
Senior Developer, Consulting Firm
“The built-in RBAC, audit logging, and GDPR tools meant we could deploy to enterprise clients without building compliance from scratch.”
Priya R.
CTO, LegalTech Company
See how teams are using it in production
Internal Knowledge Base
Deployed for 500+ employees to search company docs
Legal Document Analysis
Law firm using it for contract Q&A with citations
Customer Support Bot
E-commerce site answering product questions from manuals
Research Assistant
Academic team searching through published papers
Built something with RAG Starter Kit? Share your project
This is not a SaaS. There is no pricing page. No vendor lock-in. No central server. Clone it, own it, deploy it. Your data never leaves your infrastructure.
Submit PRs for features, fixes, or improvements. All contributions welcome.
See Contributing Guide