n8n vs node-red Daniel Payne 3 min read · 1 day ago 1 day ago -- Listen Share
Which to use
Photo by Stackie Jia on Unsplash
TLDR
Try both, but my personal opinion is:
n8n if you are using public cloud services for your data. But you will need to use something else if you want to visualize your information, e.g. ObservableHQ.
node-red if you are using private sources, especially file based data or sensor data. But you will need to use something else for AI intensive workloads, e.g. FlowiseAI.
Reasoning
There are only basically two purposes when creating commercial computer systems;
Improve decision making; by turning data into information.
and
Reduce Boredom; by spiking dopamine or releasing serotonin.
There is a third, produce regulatory data. But that’s very boring and you don’t want to read about that.
Decision Making
I don’t know very much about reducing boredom, so I will concentrate on how these two tools can improving decision making.
Both products are focused on turning data into information, and sometimes that information is a trigger for another system to do something. They both excel at this task, however each lack one main component that is needed for effective transformations.
n8n is missing a UI
In my experience you can’t just create a computer system and leave it to run. That way lies disaster, all systems need “baby sitting”. Even in transforming data, we need to see intermediate steps, progress information and summaries of results.
node-red has Dashboard 2, an excellent visualization library built around vueifity and chartJS.
n8n is missing this, so I use ObservableHQ as my go-to visualization solution, but it can be a bit of a kludge accessing live logs.
node-red is missing Antigenic AI
These days we need to use LLMs to transform data, as well as extracting insights. node-red does have some LLM nodes, but it is missing integration with LangGraph, the main tool that is used to create Antigenic workflows.
n8n is fully integrated into this landscape, and has excellent nodes for Antigenic AI.
node-red is missing this, so I use Flowise AI as my platform for doing all things Antigenic. But I use node-red to prepare my data and put it into storage systems that Flowise can use. node-red is good at ingesting data into relational databases, creating vectors (with it’s basic LLM nodes) for storage, and also processing data ready for ingestion into graph data-stores. The Dashboard 2 visualization is very important in ensuring things don’t go wrong; bad data in = bad information out.
AI Ethics
If you are new to AI and just starting out, AI ethics wont mean much, but it is going to be very important once you start building commercial systems.
The issue with AI is that it is
LLMs are non-deterministic; that is each time you run something you might get a different answer.
When you make decisions based on AI, you should always record the data & process that lead to the decision, document and log everything, especially in production.
As time goes by you will need to analyse all the bad decisions your system has made, so that you can improve future performance. You might also have to compensate prior consumers if they are effected by bad decisions.
Both n8n and node-red have good tools to record to persistent storage. However self-hosted n8n has the advantage here, as it has integration with LangSmith, a tool for monitoring AI.
Projects referenced in the article