Who maintains our automations when they break?

Whoever built it — and if they have left or are too busy, you do not have an automation, you have a liability nobody is watching.

Start from this question

Free · a few minutes · starts from “Something in the team isn't working”

The short answer

Whoever built it. If that person has left, or is too busy to look at it, you do not have an automation — you have a liability that nobody is watching. Before building anything, name the owner and decide how you would notice a silent failure, because the dangerous case is not the one that stops, it is the one that keeps running and quietly does the wrong thing. If nobody internal can maintain it, you are either buying a supported product or hiring an agency permanently.

Everything written about automation is written by people selling the building of it, so the entire category discusses the build and almost none of it discusses the twenty-three months afterwards. Search this question and you will mostly find software testing content, because the business-operations version of it is barely served at all.

That absence matters because maintenance ownership is the criterion the category omits, and it is the one that decides whether an automation is an asset or a slow-moving problem. A workflow is a small piece of software. It has dependencies that change without asking, and it will outlive somebody's employment.

Is it really this?

Worth checking before you spend anything. These are the signals that tend to tell the two apart.

Signs it is

Something is running that nobody in the business could open and explain.
You found out it had failed because a person noticed a missing outcome, not because anything told you.
The person who built it has left, or has not looked at it since they built it.

Signs it's something else

Each automation has a named owner who would notice and could fix it. That is a maintained system.
The breakage is really a process that changed underneath, and the automation is correctly reflecting a decision nobody recorded.
Nothing is automated yet, and this is a question about whether to start — which is a different page.

What's usually behind it

A workflow built by somebody enthusiastic who has since changed jobs or moved on to other work.
An agency engagement that ended at handover, with no plan for what happens after it.
A platform change that broke something nobody had touched in months.

What people usually get wrong

Treating an automation as finished at handover. It is the start of an obligation, not the end of a project.
Monitoring for failure and not for wrongness. Silent success with bad output is the expensive case.
Assuming an agency relationship ends when the build does. If nobody internal can maintain it, that relationship is permanent whether it was priced that way or not.

The options, and when each one fits

In no particular order. Which of these fits depends on your situation, not on the question you typed.

Name an owner for each automation
Always, and before the next one is built.
Switch it off
Nobody can maintain it, nobody can explain it, and the manual version is survivable.
Replace it with a supported product
Something available does the same job with a vendor responsible for keeping it working.
Build the capability internally
Automation is becoming central enough that somebody needs to genuinely own the estate.

When doing nothing is the right call

Where an automation saves a modest amount of time and its failure is obvious and harmless, leaving it alone is reasonable. What is not reasonable is not knowing which of your automations that describes.

What here comes from somewhere else

The reasoning above is ours and does not go out of date. These do: they are other people's facts, and they are linked rather than restated so you can see the current version rather than our copy of it. Last checked 16 September 2026.

API deprecations — OpenAI (opens in a new tab)That platform capabilities are withdrawn on published timetables, and that something built on one will eventually need changing. Vendors publish these notices; the decision logic on this page does not depend on any particular one being current.
Migrating from Agents to AI by Zapier — Zapier (opens in a new tab)A worked example of the same thing from a different vendor: a product being folded into another, with a migration for anybody who built on it.

What the diagnostic asks

What is going wrong and how long it has been happening.
Who owns this work today and how much of it is written down.
How repeatable the process is and how often it runs.
What a mistake costs and how sensitive the information is.

“I don't know” is always an answer. Nothing is assumed on your behalf.

The routes it weighs

Every diagnostic weighs all ten routes. These are the ones that tend to matter for this question — a description of the rules, not a promise about your answer.

Process improvement
Change how the work flows — remove steps, name an owner, fix the hand-offs.
Do nothing yet
Record the issue, set a revisit date, and decide what would change your mind.
Specialist support
Bring in a freelancer, consultant or specialist for a defined piece of work.
Develop
Build the capability in someone already in the team, with time and support.
Outsource
Hand a defined, ongoing scope of work to an external provider.
AI Agent
Delegate a defined, repeatable workflow to software inside boundaries you set, with human oversight.
AI Assist
Keep a person in charge and give them AI for drafting, summarising, checking or analysis.
Redistribute
Move ownership of the work to where it fits, and make that explicit.

What changes the answer

Nobody able to name the owner is the finding, and it usually matters more than whatever broke.
An automation touching money or personal data raises the cost of a silent failure substantially.
A supported product doing the same job puts a vendor on the hook instead of you.

Get your own read

A handful of short questions, then the strongest route for your situation, what's also credible, what we still don't know and what would change our view. No account, no setup.