Exploring Javascript Decorators

Javascript decorators are currently a stage 3 proposal, meaning they will soon be implemented in Javascript engines. But decorators can already be used today (with Babel or Typescript), and many of you are probably already familiar with them from frameworks such as Angular or MobX. They enable library authors to build new kinds of APIs, and in this talk, we're gonna do a deep dive into the current proposal, how you can use them in your libraries, and what to look out for. There will be lots of examples that you will be able to apply directly after the talk.

Building AI-powered applications with Langchain

AI technologies, such as OpenAIs GPT-4, have taken the world by storm, unlocking an array of innovative applications across various domains. In this talk, we will explore how we can use the popular Langchain framework to build AI-powered (web) applications.

We will look at a few popular use cases, and I will show you how they can be implemented. The goal is that by the end of the talk, you can go back to work and put what you've learned to practice on your own.

I have a few talks about developing AI-powered applications, and this is the most basic one. It gives a great overview of the topic and is mainly targeted at web developers who don't have much experience with AI yet.

Take your RAG to the Max!

Large Language Models are revolutionizing the world right now, but they have significant flaws: their knowledge is not up-to-date, they don't know about your company's specific data, and they start to hallucinate when they don't know the answer to a question.
An often-used technique to counter those points is Retrieval Augmented Generation (RAG), which augments prompts to the LLM with specific knowledge - often fetched from a Vector Database. In this talk, we will explore the steps needed to develop your own RAG system and then apply various steps to make it highly efficient.

This talk is one of several I have on the topic of developing AI-powered applications. It's a bit more advanced and focuses solely on building a highly efficient RAG system. It's geared towards an audience of developers with (at least a little) experience in building AI-powered applications (e.g., with the OpenAI API).

AI and the Democratization of Software Development

Artificial intelligence not only improves the lives of software developers with tools like Github's Copilot or ChatGPT, but it's also revolutionizing software development with a new generation of no-code tools like Sutro. With Sutro, users can build complete apps from just a single prompt, enabling completely non-technical people to build applications. Join us to learn how AI is fostering a more inclusive and diverse community of creators, unlocking a world of new opportunities, and driving the future of software development.

Harness Engineering in Practice

When you use Claude Code or Codex, the model itself is only one piece. The rest — the execution loop, context management, error recovery, and session persistence — is the harness. It's the layer of code that sits between you and the LLM, turning a stateless model into something that can work on real tasks across many turns and tool calls. It's what makes Claude Code so incredibly powerful.

Developers use agent harnesses every day, but most treat them as black boxes. You type a prompt, tools run, code appears. What happens between your message and the model's response is more complex and more fragile than people assume.

This talk opens up that systems layer. We'll look at how today's popular harnesses solve the same hard problems: keeping conversations alive when they outgrow the context window, recovering from API failures mid-stream, isolating sub-agent work so the parent doesn't drown in noise, and persisting state across sessions that span hours or days.

The talk includes all the learnings from analyzing the source code of popular harnesses, as well as building OpenHarness, an open-source SDK for building powerful agent harnesses.