It's easy to forget what the creators of React knew from the start: that React is not only for building web pages. Most of us have heard of React Native, that renders to native mobile components, but React doesn't have to render to "components" at all. There's React PDF to generate PDFs and Ink for building UI in the terminal.At my company, we've built a way for developers to use React to build plugins for our webapp that render to our custom components, giving us full control over the UI design (the "how"), whilst giving the plugin developer full control UX (the "what").In my talk, I'll explain what it's like to build a React renderer and reconciler, what they are and how to use them.I'd like to do a live demo showing how React can be used to convert html to markdown. And then I'd like to demonstrate how React can render to the Real World, by controlling IOT devices.

In the TV series Knight Rider, the concept of an intelligent car was pure sci-fi. Over 40 years later, we now have the tools to bring it to life - purely in the browser. This talk explores combining AI techniques like RAG and function calling with advanced browser APIs (e.g., WebBluetooth) to create a browser-based AI agent with personality, memory, and autonomy, powered by WebAssembly and WebGPU - fully offline and privacy-preserving!

Next.js applications of today struggle with an architectural contradiction: entire routes must be either fully static or fully dynamic, forcing developers to sacrifice performance for freshness or vice versa. This all or nothing approach leads to redundant computation, increased costs, and slower user experiences.In this session, we'll explore Cache Components — an experimental Next.js feature that introduces function-level caching through the `'use cache'` directive. Building on React's streaming capabilities and Partial Prerendering infrastructure, we'll demonstrate how to achieve granular control over what gets precomputed into static shells versus what streams dynamically. You'll learn practical patterns for common scenarios like expensive computations, real-time data requirements, and the sophisticated middle ground between them.

Node.js was announced in 2009 as a single-threaded JavaScript runtime. In 2018, it became multi-threaded, and no one noticed.What can worker_threads do? What are they useful for? What's the best way of communicating between two threads?

Hit testing is easy, right? Learn how tldraw's canvas editor decides which shape is beneath the user's cursor. This talk is a deep dive into five hundred of the most confusing lines of code in tldraw, the infinite canvas SDK.

Handling async operations in UI components can be tricky – flickering pending states, inefficient state updates, and unstable user experiences are common issues. Building accessible UI components presents additional challenges, as meeting accessibility requirements often requires extra effort, which can add complexity to the development process.In the context of the Next.js App Router, this session will demonstrate how React Server Components streamline data fetching, while other React 19 tools help manage mutations, loading states, and optimistic updates more effectively. We’ll combine these with Ariakit to simplify the process of building interactive, accessible UI components with less boilerplate. You’ll leave with practical strategies for managing async operations while ensuring a smooth, reliable user experience.

As AI transforms software development, teams face a critical choice: lead the change, or risk being left behind! At Typeform, we're proactively encouraging the use of AI tools, and helping everyone level-up their AI proficiency.This means lots of things including the flexibility to try different tools, democratising access to (but controlling our expenditure on) LLMs, and ensuring we're hiring the next generation of engineers who have AI as part of their DNA.In this talk, I'll provide some practical examples of how we're achieving these things at Typeform, as well as how we're enhancing the way we collaborate with people outside of engineering using AI tools.I'll share real examples using AWS Bedrock and MCP servers, plus hard-won lessons from our journey. If your team is on the path to AI transformation, this talk will give you some practical strategies for making it a success.

There are many ways to manage state in modern React applications, but the URL itself is a powerful hidden state manager that’s often overlooked.This talk introduces `nuqs`, a library that simplifies URL state management in React. We’ll explore the benefits this approach: shareability, history navigation, and declarative client-server interaction.Through a live demo, you’ll learn how to migrate from React.useState to nuqs’ useQueryState while we navigate concepts like client-only vs server-aware search parameters. We’ll also cover good practices for type-safe, pretty URLs, and how to avoid common pitfalls such as browser rate limits and URL size constraints.If you’ve ever wondered what state belongs in the URL, and how to scale it, this talk will give you the answers.

This talk explores how to effectively run React Server Components (RSC) in highly available Kubernetes clusters. It covers the shift toward a Full Stack Experience in modern React development, where components interact directly with databases, caching layers, and messaging systems. You will learn practical strategies for caching dynamic data, managing server actions, and deploying RSC at scale with Kubernetes. The session also includes real-world lessons on avoiding common pitfalls.

The original Final Fantasy IX was a masterclass in creative constraint – pushing the PlayStation’s hardware to its limits with tricks like pre-rendered backgrounds and low-res models. In this talk, Joe Hart recreates the magic in React, diving into WebGL, canvas positioning, and CSS hacks that should probably be illegal. It’s part love letter to retro games, part deep dive into browser APIs, and part performance art.