---
title: "When the First Check Flagged Records That Were Never Broken"
canonical: https://dxdev.com/everyday-ai/before-automation-make-sure-information-is-true/
datePublished: 2026-06-27
---
A person cannot get access, or a page is not showing what it should, and the complaint looks simple enough to fix in minutes. The instinct is to fix the visible problem quickly. That is reasonable, and it is also how a small mistake turns into a larger one.

In one system, our first check flagged several records as wrong. The obvious next step was a bulk cleanup. A closer look showed that most of them were [valid records in an unusual but intentional state](/blog/most-bugs-are-dirty-data-so-build-the-diagnostic-first/). Only a smaller set needed a correction.

Had we treated that first result as the answer, the cleanup would have changed perfectly valid content. The quickest-looking fix would have edited records before anyone had checked whether the unusual case was actually an error.

## Unusual and wrong are different things

Most businesses already have plenty of information. Spreadsheets, forms, customer records, task lists and reports all tell a story. The trouble is that a surprising value is not always a bad value.

An empty field might mean a record is incomplete. It might also mean the field does not apply. A person without access might be missing an email address. Or the system might be using the right address for the wrong person. A report might look inconsistent because data is damaged, or because it is showing two legitimate parts of the process that were never meant to match.

The difference matters because automation is fast. If you tell an automated process to clean up everything that looks unusual, it can make a confident mistake much faster than a person can.

## What AI is good for here

AI earns its place when it helps a person ask better questions before acting. It can compare records, group similar cases, describe inconsistencies in plain language, and prepare a short list of items worth checking. A pile of confusing information becomes a small set of decisions a person can actually review. It does not hand over the answer, and it should not make the correction or treat a pattern as proof.

A good check, one that only looks and changes nothing, should make three things clear:

1. What looks unusual.
2. What the system believes that unusual result means.
3. What will change if someone approves a fix.

That third part is the one that matters most. In our case, a preview made the difference visible before anything changed. It showed what would be preserved and what would be corrected. The preview made the process safer, and it made the decision easier to trust.

## Try it on your own work

Think about a process that currently depends on someone scrolling through a spreadsheet, searching an inbox, or checking several systems before they can decide what is wrong.

The first AI-assisted improvement may not be an automatic fix at all. It may be a [look-only report](/everyday-ai/read-before-write/) that answers one question: **what is actually different here, and what would happen if we changed it?**

A report like that removes guesswork without pretending that judgment is unnecessary. It gives people a clearer starting point, reduces the chance of a rushed cleanup, and makes the work easier to hand off.

## Keep a person on the approval

So build the look-only report before you automate any correction. Have it separate a real error from an unusual but valid record. Then let a person inspect that evidence, decide whether a correction is justified, and approve any change that matters. A process gets safer when it can tell those two cases apart, and when someone reviews the result before anything is touched.