Low-code tools have earned their place. A form, a workflow and a dashboard that would have taken a team two months can be assembled in days. The question is never whether they work; it is whether the thing you are building stays inside their assumptions.
Good fits
- Internal processes with clear steps and modest volume: approvals, requests, simple case management.
- Anything you expect to replace within two years, where speed matters more than longevity.
- Prototypes that let you learn what the real requirements are before committing engineering time.
Where the ceiling appears
Three places, reliably: unusual data models, integrations the vendor did not anticipate, and performance at volumes above the platform's design point. Each is fine alone. Together they produce a system that mostly works and cannot be fixed, because the constraint is in someone else's runtime.
Reducing the exit cost
Keep the data somewhere you control. A low-code front end over a database you own is a replaceable component. A low-code platform holding the only copy of your operational history is a dependency you will pay to leave.