We use Make, Zapier, n8n and Airtable when they fit. We write code when the workflow needs reliability, scale, branching logic, AI guardrails or ownership.
PWBy PlugWheel
Cover image
No-code versus code is the wrong fight. The better question is: what does this workflow need to be trustworthy? Sometimes the right answer is a Zapier in an afternoon. Sometimes it is a tested Python service, a database and a dashboard. Often it is both.
Use no-code when...
The workflow is simple, linear and low-risk.
The connectors already exist and the data shape is stable.
Volume is low enough that manual review can catch the rare issue.
Speed matters more than long-term customization.
Use custom code when...
The workflow has branching logic, retries, permissions or complex validation.
A silent failure would cost money, trust or customer experience.
AI is making decisions that need guardrails, evals and logs.
You need a real dashboard, internal tool, API or owned data model.
The useful middle
Many of our best builds use no-code for fast integration surfaces and custom code for the reliable core. n8n might orchestrate the flow, Postgres might hold state, an AI agent might classify input, and a custom dashboard might give the team control.
The stack should serve the workflow. Not the other way around.
