Is your workday all Ctrl+C, Ctrl+V? AI can help

How to use artificial intelligence to ease repetitive administrative tasks and make room for work that needs your attention.

You open an email, download the attachment, hunt down three pieces of information, copy them into a spreadsheet, send almost the same reply you sent yesterday, and update another tracking sheet.

By the time you finish, five more emails have arrived. Your coffee has gone cold, and the work that needed your full concentration is still waiting.

If you work in administration, that probably sounds familiar. Some days, it feels as though we were hired to move information from one screen to another.

This is where artificial intelligence can offer something close to superpowers: help with organizing information, drafting replies, and cutting down on manual steps. Small things, repeated over and over, that can free up room in your workday.

What the research can already tell us

In an experiment published in Science (2023), 453 professionals completed writing tasks. With ChatGPT, the average time spent fell by 40%, and quality increased by 18%. Those results apply to the activities studied, rather than the entire working day. [1]

In a UK government pilot, survey respondents reported saving an average of 26 minutes a day with Microsoft 365 Copilot. Published in 2025, the evaluation used participants' estimates; it did not time those savings. [2]

Another UK report found small time savings on specific tasks but could not demonstrate an overall improvement in productivity. The report itself notes that it collected limited data to assess that effect. [3]

My take: there are good reasons to give it a try. The benefit needs to show up in your own process, once preparation, review, and corrections are included.

Four places to start

The examples below are potential uses to test in your routine. Results depend on the tool, the documents, and how clearly you explain what you need.

1. That email you write every week. With confirmed information and an approved template, AI can prepare a first draft, adjust the wording for the recipient, and organize the next steps. You check the content and send it. This could be a polite payment reminder, instructions about required documents, or a progress update.

2. The meeting that creates another task once it is over. Using your notes or an authorized transcript, ask for draft minutes listing decisions, owners, deadlines, and outstanding items. Where information is missing, tell it to write “not provided.” The person who stayed quiet during the meeting will appreciate not being assigned a task nobody agreed on.

3. The document where you only need to find a few details. Request a table with the fields you need and the source passage or page for each piece of information. This can help you compare proposals or organize requests. Check the references against the original: an attractive table can still contain a misinterpreted field.

4. The spreadsheet formula you always forget. Explain which columns you have, provide fictional data, and describe the result you expect. Ask for a formula and an explanation of how it works. Then test cases you can solve yourself. For sums and calculations, rely on the spreadsheet's calculation engine and check the result.

When the rule is fixed, a formula, an import, or conventional automation may handle the repetitive work. AI can also help you build that solution. Before anything else, it is worth asking whether that step still needs to exist.

A prompt you can adapt today

“Organize this” leaves a lot open to interpretation. A more useful request explains the material, the deliverable, and the boundaries. To organize meeting notes, for example:

I will provide notes from a meeting. Use only the information they contain. Create a table with these columns: topic, decision, owner, deadline, and outstanding item. Distinguish suggestions from approved decisions. For each row, include the passage that supports the information. If a field is missing, write “not provided.” At the end, list questions I need to clarify before finalizing the minutes. Do not assign tasks or create dates on your own.

This prompt helps guide the response. Checking it is still part of the work, including verifying that the tool followed the instructions.

Once you have refined the result, save the prompt template and an approved example. Next time, you can start with a process you already know.

How to tell whether you actually saved time

Choose one frequent task whose result is easy to check. For a week, compare similar instances completed with and without AI assistance.

Record the whole process: preparing the data, explaining the request, checking, correcting, and finishing. Also watch for errors, missing information, and extra work passed on to colleagues.

Consider a hypothetical example: preparing a report used to take 30 minutes. With AI, you spend ten minutes preparing the material and generating the first draft, plus five minutes reviewing it. You saved 15 minutes. Repeated five times a week, that would give you 75 minutes back.

If corrections eat up all the time you saved, adjust the process or choose another task. A quick response only helps when the final delivery becomes easier too.

In a team, the test works better with a shared template, clear quality criteria, and someone responsible for reviewing the result. That way, everyone can learn from it.

Knowing the work still makes a difference

NIST identifies the generation of false information presented with confidence and the exposure of data among the risks of generative AI. [4] Start with fictional examples and use tools authorized by your organization when handling work information.

When reviewing, pay particular attention to names, amounts, dates, and decisions. Minutes that invent a deadline create work for other people.

Someone who knows the process can spot an odd detail, recognize that another person needs to be heard, or identify an exception that calls for care. That knowledge helps guide the tool and assess what it produces.

To me, the best use of this “superpower” is to make room for better service, resolving outstanding issues, and thinking about improvements. The team also needs to agree on how to use the time recovered, so that greater efficiency brings real relief to the working day.

Start with the task that makes you think, “This again?” It could be a good place to experiment. And, with a little luck, your next coffee will still be hot.

Sources

[1] Noy and Zhang. Experimental evidence on the productivity effects of generative artificial intelligence. Science, 2023.

[2] Government Digital Service. Microsoft 365 Copilot Experiment: Cross-Government Findings Report. June 2025.

[3] Department for Business and Trade. The Evaluation of the M365 Copilot Pilot. August 2025.

[4] NIST. Artificial Intelligence Risk Management Framework: Generative Artificial Intelligence Profile. July 2024.