ReactNext 2026

AI is reshaping frontend development faster than any shift we have seen in years. But this talk is not about hype, and it is not another “AI will replace developers” narrative. It is about how frontend engineers can stay relevant, effective, and responsible in an AI-first workflow. We will cover how to choose the right tools in a noisy ecosystem, how to move from “code writer” to “technical judge,” and how to keep professional standards high when AI generates code at scale. I will share practical guidance on what still requires human ownership: UX quality, performance, accessibility, compatibility, and maintainability.

Add to Calendarcalendarcalendarcalendarcalendar

Claude Code, GitHub Copilot CLI, OpenAI Codex CLI, Prisma, Gatsby. What do they all have in common? They're React apps - just not in your browser.

Ink is a React renderer that outputs to the terminal instead of the DOM. It uses the same Yoga flexbox engine as React Native, supports hooks and context, and lets you compose complex, interactive terminal UIs using the exact mental model you already have.

In this live coding session, we'll start from an empty index.tsx and build "Troll Code": a working AI-powered CLI, live!

Compose terminal layouts with React components and Flexbox

Manage streaming responses with useState and useEffect

Handle keyboard navigation and real user input

Push the renderer from ~30fps to optimized 60+ 

You'll leave with a working CLI, a new appreciation for React's renderer abstraction, and a clear picture of when to reach for this in real projects.

If you've ever wanted to build a dev tool, an AI assistant, or an interactive automation script, and you already know React, you're already most of the way there.

Add to Calendarcalendarcalendarcalendarcalendar

Face it, accessibility is critical for a good user experience. We are all aware of it, but we rarely treat it with the importance it deserves. Too often, it’s treated as a second-grade consideration before a release or an afterthought as enhancement. Many blame this on the complexity of compliance guidelines and a lack of clear processes. Theoretically, this is where AI shines best, due to its nature of following guidelines and best practices. Or is it? Does it mean you now can simply outsource Accessibility (a11y) to AI and stop worrying about it? What does compliance look like in an AI-assist workflow? What are the hidden cost, and how can you truly leverage these AI tools that claims to solve a11y for you? That’s what we are going to find out in my talk.

Add to Calendarcalendarcalendarcalendarcalendar

In this session, we’ll explore how Figma MCP can help create a more automatic workflow for generating UI. When combined with tools like Code Connect and VSCode Copilot, it helps teams move faster while keeping the design consistent. We’ll show how to install and use Figma MCP Desktop and Remote, and how they connect design files directly to development tools. You’ll also learn how to: - Use MCP tools to read designs and generate UI - Debug MCP calls and understand the design data returned by the system - Work with tokens, spacing, typography, and components - Extract screenshots and context from Figma A key idea is agent handoff. When different AI agents have the right skills and pass work between them, they can turn design data into working UI almost automatically. Finally, we’ll look at ways to scale this workflow for large projects by breaking designs into smaller parts. This lets teams build interfaces step by step—almost like building with LEGO."

Add to Calendarcalendarcalendarcalendarcalendar

The web as we know it is splitting. While we obsess over React Server Components, a parallel "Agentic Web" is being built. With Google’s WebMCP, Stripe's agent wallets, and Cloudflare’s "Markdown for Agents," AI bots are navigating, executing, and purchasing without ever seeing our DOM. At the same time, Generative UI is flipping the frontend paradigm, rendering complex, ephemeral React interfaces on the fly. In this session, we’ll explore how to architect React applications for this dual reality. I’ll demonstrate how to expose your app to headless agents using the Model Context Protocol (MCP), and how to safely orchestrate Generative UI components when the AI is the one writing the code. We'll look at real production patterns to ensure your frontend survives Web 4.0.

Add to Calendarcalendarcalendarcalendarcalendar

Influencers have convinced us that the answer to React’s problems are… Svelte or Solid or Vue or [insert-your-favorite-new-framework]. But are they? Don’t they all suffer from the same problems - large bundle sizes that need to be downloaded to the client, the need to manage state and sync it with the server, and others. This is because, fundamentally, they are all client-side applications, where all the code resides on the client. A new breed of frontend frameworks is proposing an alternative. These frameworks - chief among them HTMX and Hotwire/Turbo - propose a different solution, harking back to the original web days, but with the added dynamism enabled by modern browsers. Join me to learn how you can write server-side apps - where most of the app code is server-generated HTML, yet has the dynamism and interactivity of client-side apps without the pains that come with running large amounts of code in the browser.

Add to Calendarcalendarcalendarcalendarcalendar

A year ago we built an agent. Then we made it agentic. Now it's a full-blown multi-agent system. Remember when products didn't change under your feet every five minutes? I do. It was nice. In this talk, we will share Qodo's journey through AI's chaos: how we moved from single prompts to multi-agent code review, architected with LangGraph and MCPs (Figma, Linear, and more). We will walk through some of our specialized review agents. Including a runtime agent that actually runs your code changes, records the UI flow, and catches UI errors and console errors on video. I'll close with a live demo and a few hard-won lessons on staying current in a market that won't slow down. "

Add to Calendarcalendarcalendarcalendarcalendar

As developers, we usually implement given UI designs. But wouldn’t it be great to gain a better understanding of the science behind a good design? In this talk, I will cover some of the cognitive psychology principles and UX laws that guide UI/UX experts in their work. By applying these principles and laws the you can improve both the user interface and user experience, resulting in a better product.

Add to Calendarcalendarcalendarcalendarcalendar

Refactoring thousands of files to break a React-Redux monolith is a daunting task. When our initial estimation for decoupling our legacy architecture hit the 8-year mark, we knew manual refactoring wasn't an option. Instead, we built an AI infrastructure engine designed to navigate the complex dependency graphs of a modern React ecosystem. In this talk, we'll explore the custom orchestration engine driving this reality. You will walk away understanding how to combine the deterministic reliability of ASTs with the creative power of LLMs to solve your own "impossible" React migrations and finally achieve a modular architecture.

Add to Calendarcalendarcalendarcalendarcalendar

When a React component crashes today, the user hits a dead end, and we get flooded with raw logs. By leveraging Chrome’s built-in AI, we can shift from passive logging to active, client-side triage. In this talk, we’ll build a "Resilient Error Boundary" that detects crashes and uses local inference to apply immediate, transparent workarounds. keeping the interface alive. We'll explore how to communicate these temporary fixes to the user honestly while generating concise root-cause summaries for your backend. You will leave with a production-ready architectural pattern to make your applications self-healing and frustration-free.

Add to Calendarcalendarcalendarcalendarcalendar

For years, web scraping has been a fragile game of reverse-engineering HTML-defined by brittle selectors, endless maintenance, and constant breakage. But as AI agents become first-class consumers of the web, traditional scraping is no longer enough. Enter WebMCP: a protocol-driven approach that allows websites to expose structured, machine-readable capabilities directly to AI agents. In this session, we’ll explore how WebMCP replaces "DOM-guessing" with explicit, secure, and resilient tools. We’ll dive into how this shift enables seamless LLM-to-Web interaction and discuss what it means for the future of web architecture in an AI-first world.

Add to Calendarcalendarcalendarcalendarcalendar

React’s new Activity component introduces a powerful idea: rendering UI without immediately committing it to the screen. This enables applications to prepare expensive or complex UI ahead of time, improving responsiveness and perceived performance. In this talk, we’ll build a clear mental model of what the Activity component does, how it differs from Suspense and transitions, and where it fits in real-world React applications. Through practical examples, we’ll explore patterns for pre-rendering UI, coordinating user interactions, and avoiding common pitfalls. You’ll leave with concrete use cases, a deeper understanding of React’s rendering behavior, and the confidence to decide when the Activity component adds real value and when it doesn’t."

Add to Calendarcalendarcalendarcalendarcalendar

This session will be to demonstrate how we use Panda-CSS to build styling tools that are custom, typesafe and opinionated to enable engineers and agents to ace intense Ui engineering tasks with ease.

Add to Calendarcalendarcalendarcalendarcalendar

Are you still prop-drilling data through five layers of React Server Components just to avoid a second database query? It’s time to meet the React cache API - the request-scoped singleton that’s changing how we architect data flow. Join me to this lightning talk as we will cut through the noise to understand how cache differs from useMemo and why it is the "secret sauce" for high-performance RSCs.

Add to Calendarcalendarcalendarcalendarcalendar

Everyone is asking whether AI is killing frontend development, but maybe that's the wrong question.

In this panel, we'll break this big question into smaller ones: What does seniority mean when AI writes code? Is it becoming easier to fake expertise? Which skills are losing value, and which are becoming more important than ever? And what does it actually take to stay relevant today?

Instead of another discussion filled with hype or fear, we'll have an honest conversation about the things many developers are already thinking about, but don't always say out loud.

The goal is simple: lower the anxiety, make sense of what's really changing, and leave with practical ideas you can apply the day after the conference.

Add to Calendarcalendarcalendarcalendarcalendar

We replaced brittle browser tests with sleek Jest + jsdom—then learned “fast” and “simple” can still be wrong. This talk shares what failed, what ultimately worked, and a practical framework for combining unit, component, and end-to-end (E2E) tests across Jest/Vitest, Testing Library, Cypress, and Playwright. Save yourself the next rewrite. 

Add to Calendarcalendarcalendarcalendarcalendar

For years, libraries have been the default way to move faster, avoid edge cases, and build reliable UI. But with modern AI tools, we can now generate custom components in minutes, tailored exactly to our product needs - no APIs, no constraints, no waiting for feature requests. In this talk, we’ll walk through a real example of replacing a popular React library with AI-generated code. We’ll compare both approaches side by side, explore where AI shines, and uncover the hidden costs that come with owning the solution yourself. This isn’t a story about AI replacing libraries - it’s about how the tradeoffs have fundamentally changed, and what that means for how we build modern applications.

Add to Calendarcalendarcalendarcalendarcalendar

A practical, code-heavy talk showing how to build runtime API contract monitoring directly into React apps. I'll live-code a fetch interceptor hook, a confidence-based schema learner inside a React context provider, and React patterns for surfacing violations — including custom error boundaries and health hooks. Ends with a live demo breaking an API on stage and watching React catch it in real time.

Add to Calendarcalendarcalendarcalendarcalendar

Testing React apps can get messy and slow, especially as they grow. In this session, I’ll show how to build autonomous testing agents that actually learn from your app, run End-to-End and regression tests automatically, and adapt to changes in your code. We’ll look at a real-world example from production, including metrics, challenges, and what worked (and what didn’t).

Add to Calendarcalendarcalendarcalendarcalendar

“Um, Actually” is a fast-paced, interactive live gameshow where panelists compete by correcting confidently wrong statements about technology, culture, and everyday “things we think we know.”
We tailored this event to ReactNext, even though we write vanilla HTML (hardcore!)
Each round presents a plausible but flawed claim. The goal isn’t trivia - it’s identifying what’s wrong, why it’s wrong, and how small assumptions lead us astray.
The format blends humor, audience participation, and sharp thinking, turning mistakes into the main attraction.
Expect laughter, arguments over definitions, and the satisfying moment when someone finally says: “Um, actually…”

Add to Calendarcalendarcalendarcalendarcalendar

What if your next feature wasn't designed by a human—but by the app itself, in real time? Generative UI is flipping the frontend paradigm on its head, moving us from static, hand-crafted screens to interfaces that are dynamically composed based on context, intent, and AI output. In this talk, we'll dive into how modern AI models can emit not just text, but live, interactive UI components—using tools like the A2UI, MCP APPS, and tool-calling to bridge the gap between language models and rendered interfaces. We'll walk through real-world examples: a chat assistant that doesn't just describe a flight—it renders a booking card; a dashboard that assembles itself based on what the user asks. You'll also learn where generative UI breaks down, how to guard against hallucinated layouts, and how to keep the developer in control without killing the magic.

Add to Calendarcalendarcalendarcalendarcalendar