I started trying out n8n for setting up automated workflows.
I tried setting up a basic HubSpot filter to collect all deals in a certain stage. Unfortunately, the API doesn’t return the names of things, just IDs so I can’t tell by looking at it what the pipeline is or the stage because those are customized to our instance of HubSpot. This would probably not be possible for a non technical person to do.
How is it different than Dify?
n8n has out of the box integrations with many B2B SaaS tools like HubSpot. It also provides triggers like cron jobs and webhooks and Dify does not.
Notes on usage:
- Quirky UI: double click to edit, triple pane layout when editing with dismiss button on the left, when to use an expression or not for a field value, tools have to be linked visually to an LLM node
- When working off of large lists, speed up the feedback loop by only executing a node once while you test it out by going to Settings -> Execute Once (don’t for get to flip it back)
- Clicking the test button doesn’t actually seem to execute the full flow
- To extract fields from a large API response, use the Edit node (horribly named)
- Tools can only be used in conjuction with a language model (e.g. SERP can’t be used to make arbitrary search queries in a workflow)
- HubSpot has a horrible API and the built-in integration in n8n does not make it any easier (e.g. you need to map IDs yourself by making multiple API calls and using a Merge node to join them)
- Scheduled tasks don’t run until you activate the workflow
Links to this note
-
What Does it Mean to Be AI Native?
With the capabilities of large language models getting more useful for real work, the pressure is on to incorporate them everywhere. I’m seeing an increase in loud voices proclaiming people and businesses must become “AI Native” if they are to survive.