A lightning walkthrough taking an existing site and turning it into searchable content using OpenAI embeddings. We'll use react server components to create an extremely snappy UX while using server-side cosine similarity functions to determine the best search results.

Join me on the RAG (Retrieval Augmented Generation) road. In this talk we start from the basics, we will go through embeddings and vector databases. Then finally i'll show you a great use case for a RAG application that has a React front end and utilizes the capabilities of generative AI.

LLMs are powerful, but at the same time, too large and resource-intensive. In this talk, you’ll learn how these models work, how they're optimized into smaller, more efficient SLMs, and why it’s a game-changer for mobile development. We’ll also explore the benefits of Edge AI, like privacy, lower latency, and offline support, and we'll walk you through how to integrate on-device conversational AI into a React Native app using ExecuTorch.

I'll share our experience building RorkAI, an AI builder for fullstack React apps without code. It's like Replit meets Webflow/Framer. We're very early stage, but built a fully working prototype in 4 weeks, and we're proud to share our experience.
Questions we'll touch:

• Our experience building Cursor-like Composer for next with Claude and O1
• What turned out to be the best fullstack framework like Next, Remix or Tanstack Start
• How to create UI like Vercel's v0
• Share our experience of using Webcontainers, Nodebox and Sandpack
• how to avoid RAG errors & improve AI reliability
• how to make AI create authentication, database migrations, use Stripe, and external APIs (OpenAI, Replicate, etc.)
• what innovations in the JS world (not including) made RorkAI actually possible Our vision is that RorkAI will help non-technical founders build lots of new SaaS businesses with AI. We don't think AI will ever replace engineers for interesting tasks, we think it will replace outsource agencies. We help founders test business ideas faster and switch to in-house teams as they grow.