How to Build AI Automations That Others Can Actually Use
AI can generate any script, but handing it off to non-technical users is still a nightmare. I found a way to use AI to build visual workflows that others can actually run and modify. Here's my breakthrough with n8n and Cursor.
The Automation Landscape Has Flipped
The old world: Zapier and similar tools were lifesavers when coding was hard. Drag-and-drop workflows for people who couldn't write code.
The new world: AI can literally prompt any script into existence. The friction of building custom software is gone. Why bother figuring out the quirks of pseudo-code platforms when I can just tell Claude "write me a script that does X, Y, and Z"?
The problem: I can now prompt any script into existence. That works great for me, but it isn't easily handed off to someone less technical and they may have trouble running it, never mind editing or taking ownership of it going forward.
My Breakthrough
I figured out an entrypoint to those visual workflows such that I can use AI to go faster but still hand off an experience that a friend or client can easily run and manage in the future.
Here's what I discovered using n8n (an open source alternative to Zapier):
- Downloaded the entire n8n GitHub repository
- Opened it in Cursor (AI code editor)
- Asked AI to create workflows using the full context
- AI generates JSON files that import directly into n8n
- Handoff-ready automations with visual interfaces
Note: Make.com uses a similar workflow format, but you'd need to find a way to get their documentation into Cursor or your LLM chat since it's not open source.
The Game-Changer
People get both the power of AI-generated logic and a user-friendly interface they can understand and modify later.
This solves the handoff problem. Instead of giving someone a Python script they can't modify, I give them a visual workflow they can actually work with.
A Real-World Example
If you're still wondering how automation could fit into your life, here's an example from my own:
Take my consulting invoicing workflow. Nobody likes invoicing, but I don't want too much time to pass between invoices or for any invoice to be a surprise. I want to invoice weekly with context for what's being billed.
My automated workflow: looks at last week's meetings in Google Calendar, identifies all meetings for a client, pulls meeting notes from Notion, then creates a draft invoice in Stripe that lists the meetings with a tight AI-generated summary of what was discussed. I review drafts, add non-meeting items, and send.
This kind of workflow is perfect for automation because it's high value to clients but isn't very complicated. An intern or assistant could do it.
AI is transforming how we work, but you don't need to navigate it alone. I help individuals, teams, and organizations cut through the noise to find AI solutions that actually drive results.
Whether you're looking to master AI fundamentals, streamline your team's workflows, or implement organization-wide AI strategies, I'll guide you to what works for your specific needs.
Ready to transform how you work? Let's chat – I'd love to explore the possibilities with you.
What's your experience with AI-generated workflows? Are you still wrestling with the deployment problem?