Okay, this might turn into a bit of a rant. I’ve already mentioned that I’m currently working on an “AI Assistants / AI Agents” course. As part of that, I’m looking at tools like Langflow, Flowise, and n8n and trying to build typical workflows (simple chatbot with history, decision-based workflows, RAG). With almost every workflow I try, I run into problems where things simply don’t work as they should—either because of bugs or because basic functionality is missing. Sometimes nodes can’t be connected even though they should be. Other times, type mismatches prevent flows from working at all, and nodes that could solve this (like type converters) have either just been removed or don’t work with the existing types anyway. And what does ChatGPT say about it?“The tool is currently in a transition phase …” That seems to apply to all of them, although I suspect n8n is still the most stable overall. On top of that, ChatGPT often gives incorrect advice, even when you explicitly ask it to rely on the current official documentation. At that point, it becomes pretty obvious that this so-called “intelligence” has clear limits. Putting AI aside for a moment: what kind of QA is this? And what do the developers of these platforms think—how are people supposed to work with these tools if even the most basic things don’t work reliably? My assumption: everyone wants to be right at the front of the AI hype. And yes, I think we’re still very early in this phase. So new versions are pushed out constantly just to keep up, regardless of stability. If I (or we) released software like this to paying customers, they’d be gone very quickly. Maybe part of the problem is that large parts of these tools are themselves built with AI, which could explain many of the bugs and incompatibilities. But even then, basic functionality should be tested properly before a release. In any case, it doesn’t reflect well on the development teams when unfinished and buggy software is released. What do you think? Is this just normal for cutting-edge software? From my perspective, the tools that will ultimately win are the ones that are solid and free of showstoppers—but maybe nobody dares to do that out of fear of falling behind.

Welcome! This category is for everyone looking for English resources on web development. Here you’ll find: - Free tutorials and starter projects - Links to my English courses (moved here step by step) - Updates & extra resources Have a look at the classroom, there will be some english courses. More tutorials will be added over time. Alpine.js Essentials HTML Forms Essentials ES6/ES.next - Learn Modern JavaScript Alpine.js Masterclass If you have any feedback or requests, feel free to comment below 🙌

In case you missed it: the React hacks don’t seem to stop. Please update to the latest version — this also applies to Next.js and everything that uses React Server Components. I’m starting to question whether RSCs are really such a good idea. What do you think? https://react.dev/blog/2025/12/11/denial-of-service-and-source-code-exposure-in-react-server-components

Brief summary of my video about useEffect (generated by NotebookLM): The React useEffect Hook is essential for side effects like data fetching. But watch out for these two common traps that can lead to broken or buggy application behavior. 1. The Missing Dependency Trap Many developers use an empty dependency array ([]) to make useEffect run only once when the component mounts. The Problem: If your effect logic uses a state variable (like a Filter), but that variable is not included in the dependency array, the effect will not rerun when the state changes. React itself cannot automatically detect this dependency. You'll see the component re-render with the new state (e.g., 'Sale'), but the product list remains unchanged (e.g., still 'Popular' products 1, 2, 3). The Fix: Always provide an array containing all dependent values used within the effect's callback. By including the state variable Filter, React notices when the value changes, and the effect is correctly triggered again to fetch the updated data. Action Item: If your effect isn't running when you expect it to, check your Dependency Array first! 2. The async/await Trap A major error occurs when you try to declare the useEffect callback function itself as async. The Problem: The useEffect callback must return either nothing (undefined) or a cleanup function. An async function, however, implicitly returns a Promise. Since a Promise is not a valid cleanup function, React throws an error: "cleanup function must be a function or undefined". The Fixes (Keep the Outer Function Synchronous): 1. Use .then(): Avoid async/await and use the Promise API syntax with .then() to handle the asynchronous result. 2. Declare an Inner Function (Recommended): Define a separate, named async function (e.g., innerFetch) inside the useEffect callback. The outer useEffect callback remains synchronous and simply calls this inner function, successfully avoiding the implicit Promise return. --------------------------------------------------------------------------------

Hey Devs, as some of you might already know, you can find my courses here in the Classroom — these are (atm) the same ones I offer on Udemy. If you’re interested, I recommend checking them out here directly — on Udemy, the prices are usually higher because I’m no longer part of the automatic discount program. Also, if a course is purchased on Udemy without one of my promo links, I only receive a small share of the sale. Here on Skool, the courses are currently cheaper (at least for now 😉) — plus, we can discuss questions or related topics directly here in the feed if you’d like to dive deeper.(That’s completely optional of course — the courses are still self-paced as usual.) Cheers, Stephan