A low-code platform lowers the barrier to development, not the obligations associated with the built system. A workflow assembled by a department in an afternoon accesses real data, executes actions on production systems, and creates an operational dependency exactly like an application developed by the technical team. The difference is that no one has reviewed its permissions, no one knows it exists, and no one will maintain it when its author moves on.
This is not an argument against low-code. It is an argument in favor of treating it as what it is: production software.
Obligation | In a formal development | In a low-code flow |
|---|---|---|
Permits | Design revised | They inherit the account from the person who created it. |
Traceability | Designed | It depends on what the platform registers |
Continuity | Documentation and relief | It lives in a person's head |
Compliance | Evaluated before deployment | Nobody has evaluated it |
The permissions queue is the most dangerous. A workflow created by a finance person is typically executed with their credentials. This means two things: the workflow can see everything that person sees, and when that person changes departments or leaves, the workflow will either stop working or continue to function with permissions it should no longer have. Both outcomes are bad.
The continuity queue is what creates emergencies. A billing process that depends on a workflow understood only by someone who is no longer there is a real operational risk, and it always appears at the worst possible time.
It's important to acknowledge the merit, because the answer can't be prohibition. Low-code thrives because it solves a genuine problem: departments have small, urgent needs, and the formal technology channel takes weeks or months to address them.
As long as that speed difference exists, people will build on their own. It's the same mechanism that produces the Shadow AI that enters through the corporate card , And it is corrected in the same way: not by closing the informal route, but by making the formal route fast enough.
The practical way to govern this without hindering anyone is to classify by consequence, not by tool:
Level 1 — Personnel. It only affects the creator; it doesn't touch sensitive data or write to shared systems. Total freedom, no paperwork.
Level 2 — Departmental. It is used by a team, reads shared data, and does not perform irreversible actions. It requires inventory registration, a designated owner, and an annual review.
Level 3 — Operational. It handles personal data, performs actions on clients or money, or some process depends on it. It requires permissions review, traceability, documentation, and a succession plan. It is production software and is treated as such.
Most of what is built falls into level 1, and stating this explicitly is important: a policy that treats everything as level 3 will be ignored, and rightly so.
To classify any flow in five minutes:
Inventory is the deliverable that makes everything else possible, and it is better achieved through amnesty than through auditing:
Ask without penalizing. A communication that explains why it is being done and guarantees that nothing will be turned off without prior discussion.
Offer something in return. Support when it breaks, help connecting it properly, backup if its author leaves. People record what they did when recording has advantages.
Review only level 3. Focus your efforts where the risk is, and leave the rest alone.
Create a fast track. A channel for requesting help with a response within days. The speed of that channel determines how much of the problem spirals out of control.
There is one question that governs this in any committee: How many of the company's processes today depend on something that was built by one person alone and that no one has reviewed?
If nobody knows the answer, that's the point of the project. It's not about banning low-code, which solves real problems, but about understanding what's out there, classifying it accordingly, and protecting what matters.
What you can't do is have software in production that the organization doesn't know it depends on.
Four obligations do not disappear even if the technical barrier is lowered: permissions, which usually inherit the personal account of the person who created the flow; traceability, which depends on what the platform records; continuity, if the knowledge lives in a single person; and regulatory compliance, which no one has evaluated.
No. It thrives because it solves a real problem: departments have small, urgent needs that the formal system takes weeks to address. As long as this speed difference exists, informal construction will continue, so the solution is to make the formal system faster, not to close the informal one.
Consequently, there are three levels: personal, if it only affects the person who created it and does not touch sensitive data; departmental, if it is used by a team and reads shared data without performing irreversible actions; and operational, if it touches personal data, moves money, or some process depends on it.
Under what credentials does it run, what data does it access, does it write to any system or only read, what would happen if it stopped working tomorrow, and who would maintain it if its author were unavailable.
With amnesty instead of audit: communicate why it is being done and ensure that nothing will be shut down without prior conversation, offer support and help in exchange for registration, review in depth only the operational level flows and create a request channel with a response in days.
Either the workflow stops working because it was running with its credentials, or it continues to run with permissions that should no longer be active. Both outcomes are problematic, which is why operational-level workflows need their own identity, documentation, and a succession plan.
Do you know which low-code workflows your operation depends on today? We take inventory, classify accordingly, and ensure permits and continuity for those that matter. Let's talk → |
Español Español 
English
Français
العربية
Português
Deutsch
Italiano
