Season · AI, Trust & Human Oversight · 14 of 16
The typing can disappear while the duplicate work stays.
Concept of the Week · Remove the repeat before automating it
Before automating a repeated task, check why it needs to be repeated. Generating the same information in several places can leave someone checking and correcting every copy. The useful measure is the work that remains, not the amount the system produces.
Design rule: Remove duplicate documentation before automating what remains.
On a general inpatient ward, a nurse records helping a patient change position so colleagues can see what care was provided and when. An AI documentation assistant turns the nurse's spoken account into several draft entries for the same event, each edited separately.
The nurse must inspect each draft and repeat the correction wherever the information was copied. That rework can keep them at the chart instead of returning to the next care task, even though they typed less.
A calmer workflow would first remove duplicate entries that serve no distinct purpose. The assistant could then draft a single record that captures the patient's identity, when the care occurred, and what the nurse said. After the nurse checks and authorizes the record, other parts of the chart could display that same record wherever appropriate. A later position change or a new assessment would still require its own entry.
One correction to the shared record could update each view and preserve the change history, saving the nurse from fixing each copy.
Ship this by removing redundant entries, then test whether nurses can record a position change once and have any correction appear in every view that uses that record.

A ward nurse reviews the same care event in two designs: separate drafts on the left, one draft for use across the chart on the right.
The assistant can fill in draft fields from speech, but the team must decide which duplicate entries to remove. It must not treat silence as evidence that care occurred or present an old observation as a new assessment. Compare the current workflow, a simpler workflow without the assistant, and the AI-assisted version. Measure the time spent speaking, checking, correcting, and entering data manually, and verify that the record is complete and accurate.
Asset & Resource
Before adding an AI assistant, design and human factors teams can use the Duplicate Documentation Review with nurses and the individuals responsible for documentation requirements. Revisit it when required fields, templates, or connected systems change.
Trace one event: Follow one completed activity through every place it is recorded.
Explain each entry: Identify who uses the entry and whether it adds necessary information not recorded elsewhere.
Remove the repeat: Agree on which repeated entries to remove and where the reviewed record can be displayed.
Define the remaining check: Show the patient's identity, when the care occurred, the source account, and anything still needing a check.
Measure all the work: Measure time spent speaking, checking, correcting, and entering data, and verify the record's accuracy and completeness.
Complete the Duplicate Documentation Review using the actual forms and workflow. Ask nurses who will use the system to test a draft with errors, an interrupted save, and a later care event under realistic working conditions. Check that errors remain visible and that later care is recorded separately. Use what you observe to improve the design, then test it again. Before release, check the revised design against agreed workload and safety criteria.
Research on nurses' documentation experiences reports repeated entry of the same findings, including positioning and activity, across different parts of the record. The study supports investigating repeated work; it does not establish whether this proposed workflow or the Duplicate Documentation Review reduces it.
Light Wisdom & Reflection
“What information consumes is rather obvious: it consumes the attention of its recipients.”
Which repeated step have we made faster without checking whether anyone still needs it?
Mindful Practice
While washing a cup at home, take a minute to notice when it feels clean.
What tells you the task is complete?
When do you find yourself repeating the same movement?
Next time, notice the difference between a useful repetition and one that simply follows the last.
Best,
Andreas Walden
Share this with someone deciding what a documentation assistant should take off people's hands.
When separate versions are necessary, revisit how to make changes between AI analyses easier to inspect.
