# Everyone wants to automate the workflow. Who owns it when the automation breaks? — retained record

> Retained reddit record from reddit:AI_Agents.

Canonical: https://highsignal.app/data/records/4fc902cea4d4915b

[← sources](https://highsignal.app/data)


source: [reddit](https://highsignal.app/data/reddit) (reddit:AI_Agents)  published 2026-11-17 06:10 ingested 2026-11-18 02:36

[original source ↗](https://www.reddit.com/r/AI_Agents/comments/1xycx96/everyone_wants_to_automate_the_workflow_who_owns/)

## retained text



```
There's a lot of excitement around automating business processes with AI.

And I understand why.

A workflow that takes an employee two hours might be reduced to a few minutes.

That's a meaningful improvement.

But there's one question I think gets overlooked.

Who is responsible when the automation makes a mistake?

Imagine an AI workflow that reads incoming invoices, extracts information, and updates your accounting system.

It works correctly most of the time.

Then one day, it processes a duplicate invoice.

The system doesn't crash. Nothing looks obviously broken. The workflow simply completes the wrong action.

Who notices?

Who investigates?

Who decides whether the automation should be paused?

This is why I think AI automation needs more than a good prompt and a working integration.

It needs ownership, monitoring, clear limits, and a way to recover when things go wrong.

The technology can automate the task.

It doesn't automatically take responsibility for the outcome.

**For those running AI automations in production, who actually owns the workflow after deployment: engineering, operations, or the business team?**
```



document fetched 2026-11-18 00:20 · canonical https://reddit.com/r/AI_Agents/comments/1xycx96/everyone_wants_to_automate_the_workflow_who_owns
