I've been a dedicated Node-RED user for years and absolutely love the ecosystem. I've used it for everything from serving as the rules engine for Home Assistant (the built-in HA automation model just doesn't click with me) to rapidly deploying services at a startup - services that eventually got replaced by more robust solutions. Whenever I needed to experiment with new technology integrations, Node-RED was always my go-to tool.
However, over the past 6 months, I've noticed a shift in my workflow. Where I used to turn to Node-RED for those quick 15-30 minute integrations, I now find myself spending 5 minutes with Claude to build a Product Requirements Document, then having Claude write the code in Go (which I don't actually program in myself). I can test it, deploy it to my Docker stack, and integrate it with my existing alerting infrastructure - all within 30 minutes.
As a result, I'm finding myself using Node-RED less than before. I'm curious if anyone else has had similar experiences, or if you've found different patterns in how your toolchain has evolved.
No, much the same for me. I use NR as the framework that lets me build workflows quickly and keeps things consistent and together. It's ability to interface between different data sources and to respond to many different events easily and consistently is key.
For me personally, splitting things into multiple microservices would result in much larger management overheads which I neither have time for nor do I find that work interesting.
For me, Node-RED presents pretty much the ideal balance between an ability to do things exactly the way I want and having everything in a single platform for ease of management and development.
It is also ideal from a language perspective as I long ago decided that I only had the time and energy to focus on a single computer language - the only logical decision being to use JavaScript throughout. (Though, of course, HTML and CSS have to be used as well).
My use of AI for compute mainly revolves around allowing me to do more complex tasks than I would otherwise have the time to work through.
absolutely , I was stuck with c/c++ what I used when is was paid to program in the early 90s and javascript because it was just so easy to pick up and so flexible (and it kind of reminded me of REXX) . Now with LLMs , I am learning x86 assembler - (have a new lesson every day ) , am using typescript when doing anything non trivial , and for things that I know one day could scale - I can use a appropriate language like go. Most importantly though as a backend person who has no front end knowledge, skills or feel I can now get simple (albeit not exciting UIs) up in minutes something I just could not do.
Still love Node Red , just finding it is now not necessarily my first goto for new things
Ah, one of my old favourites. Learned it on IBM mainframes and was delighted to find that it was integral to Commadore Amiga's.
I've not been an actual paid professional developer for a very long time now. When I was, I picked up PHP initially along with Python for doing web programming. When I stopped being a dev, I needed to regroup and minimise the languages I was using. So JavaScript with HTML and CSS was it. Along with the occasional VBA, VB for Office'y type stuff and later, C/C++ (though only a fairly noddy level of knowledge) when I started to get into IoT for home automation.
When I started work on UIBUILDER, around a decade ago now! I had to double-down on JS/HTML/CSS so I now have even less time and energy to use other languages. Thankfully, during that time, the need to use VBA dropped right off for me and my use of the Arduino IDE also dropped off once I discovered ESPHome.
Of course, also during that time, the pace of change for JS, HTML and CSS has picked up enormously. So never a shortage of new things to learn. AI certainly helps in my learning. Even when the AI gets something wrong, I may well learn something new from it.
For me Node-RED has been a fantastic upgrade and replacement of my earlier home automation system that was based on EventGhost (r.i.p.) and kept the automation running for almost 10 year
The principal and personal guidance for me when setting up a home automation is "set, forget, do your stuff, repair yourself if needed & notify me about everything important that might happen". So far this has worked so very well with Node-RED, so easy to visualy configure all kind of rules and exception handling
In terms of AI, I guess in my case where it already kicks in, is in the video analytics I use. I have the latest YOLOv11 running on a separate platform (NVIDIA Jetson Orin) that communicates the results back to Node-RED. So when new AI video analytics evolves I will with high interest continue to follow and try to keep up with the latest & greatest
But I will not add features that allows my system to listen to what we say or talk about
In addition to Node-RED itself, I love Python and I have a ton of external services running, doin there stuff, all communicating with Node-RED via MQTT (love it too)
No so much. I use Gemini AI Studio somewhat to write code for using various npm modules in the function node.
I cannot say that it is better than looking at the documentation or on stackexchange. It makes simple errors doing basic tasks on hugely popular npm modules like tesseract.js (50K downloads per day).
This means I have to read the documentation or paste into VS code so I can use its autosense completion to find and fix the error. Man I really wish Node-RED could do autosense completion.
I guess the reports of LLM for programming tasks 20% faster but actually 19% slower is very true.
just used n8n for the first time - the builtin LLM over the documentation is a very powerful feature for helping new users to get up to speed- I went from zero to a flow receiving e-mail from gmail with optional attachments - parsing the attachment , combining into single user prompt - adding system prompt and calling Ollama in 30 minutes. a pretty simple flow and the effort is very similar in nr however the llm over the documentation made it significantly less intimidating. every time I had a question I went to n8n Docs AI and it got very close to the soln - food for thought
I am not sure what it is about n8n, but I am enjoying it despite not liking it anywhere as much as node-red. There are some nice parts to it. Time will tell if I stick to it.
I'm currently working on an automatic RSS podcast creator.
Using multiple LLMs + google text to speech + reverse proxy + google cloud to scrape blogs for new posts and auto-create RRS podcasts with a short and long summary, TTS of the original post + link to the original post.
This way I'll have automated access to the latest blogs from sites that I like. So far, so good!
)