WebBluetooth – the Missing Link

JS Nation

JSNation 2022

Nico Martin

useMachineLearning… and Have Fun with It!

React Summit

React Summit 2023

Superwebapps: Rethinking Desktop Applications With Progressive Web Apps

React Summit 2024

Let’s Build K.I.T.T. With JavaScript

JSNation 2025

Getting Started With AI (-Agents) in the Browser

React Advanced

React Advanced 2025

Let’s build K.I.T.T. With JavaScript

JSNation US

JSNation US 2026

In this talk, we’ll explore how to build multimodal AI agents that understand voice, interpret images, hold short-term memory, and respond intelligently - all running entirely in the browser. No server, no cloud APIs, and no installation. Using modern tools like WebGPU and WebAssembly, we’ll see how far the open web has come in enabling real-time, privacy-preserving AI experiences.

Look Ma, No Hands! Multimodal AI-Agents in the Browser

AI Coding Summit

AI Coding Summit Berlin

Browser-based AI inference is stuck between two bad options: follow a standard and get portability at the cost of speed, or hand-tune your own kernels and burn weeks of deep GPU expertise per model. For Transformers.js, we tried something that probably shouldn't have worked as well as it did: handing that hand-tuning work to an AI coding agent, with way less supervision than seemed responsible. It found multiple-x speedups across LLMs, audio transcription and text-embedding models running in the browser, on hardware we already had. This talk is a case study, not a pitch. I'll walk through the loop that turned autonomous kernel optimization from chaotic into trustworthy, the one instruction that changed the agent's behavior more than any amount of prompt engineering, and the counterintuitive lessons from real experiments, including why our biggest wins did not come from clever algorithms and why you should never let your agent gaslight you into thinking it's done.

"I'm AFK for a Bit": Letting an AI Agent Run Wild on WebGPU Kernels

React Day Berlin

React Day Berlin 2026

Client-side AI is everywhere. With Transformers.js and WebGPU, we can run ML models directly in the browser.. But should we? In this talk, we'll explore when client-side AI makes sense and when it doesn't. We'll cover the technical foundations, examine real-world use cases, and compare the trade-offs: privacy vs. performance, small models vs. big capabilities, client vs. server. Which tasks belong in the browser, and which belong in the cloud? It depends!

It Depends: The Final and Honest Take On Client-Side AI

In this workshop, we will build our first AI Agent – completely from scratch. We will cover the fundamentals of LLMs, explore Retrieval-Augmented Generation (RAG), and learn how to use function calling to connect AI with external tools.By the end, you will have a working AI agent that can autonomously decide which functions and data sources to use. No prior AI experience is required, but basic knowledge of TypeScript is recommended.After this workshop, you will be able to...<ul><li>Understand how Large Language Models (LLMs) work and how they process and generate text.</li><li>Leverage LLMs for specific tasks by designing prompts and integrating them into applications.</li><li>Implement Retrieval-Augmented Generation (RAG) to enhance AI responses with external knowledge.</li><li>Use function calling in AI systems to interact dynamically with APIs and data sources.</li><li>Build an AI agent that autonomously selects the right tools and data to complete user tasks efficiently</li></ul>Who is this workshop for?No prior knowledge of Artificial Intelligence is necessary, as we’ll cover all the essential concepts during the workshop.We’ll build the AI agent in a React app, but the core logic will be framework-agnostic and written in TypeScript.

React Summit US

React Summit US 2026

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!

Progressive web apps describe a whole new set of browser features that enable completely new ways of interacting with web applications.While originally intended as an alternative to mobile apps, we are seeing more and more desktop apps that rely on the web platform. Photoshop, Visual Studio Code, Squoosh, Clipchamp and the whole Google Office Suite are just a few examples of a growing number of completely web-based desktop applications - Superwebapps!.In this talk, we will focus on some of the new and exciting browser capabilities that allow us to create powerful React applications that make our daily lives easier. Of course, including tangible examples and an outlook on the browser roadmap.

Machine learning is seen by many as the next step in artificial intelligence towards a new stage of human evolution. And thus helps us find new approaches to solving real-world problems. Phew... That sounds complex… And how is that supposed to be fun?
Well, in addition to the big issues of our time, it is ultimately just another tool that we can play with. While it is important to first understand the core concepts of machine learning, we can quickly go way beyond that. Get ready for some unexpected examples of how to get started with machine learning in your React application!

The WebBluetooth API finally closes the gap between the browser and devices around us. And that suddenly opens up a rabbit hole, in which we inevitably encounter a few hurdles - the story of a frontend developer dipping his toes into the IOT world.