AI at work: where to start

Every team is being told to use AI, and most do not know where to begin. The answer is not a tool. It is a task. Choose a piece of work you already do, learn how an AI assistant handles it, and keep it only if it saves time without lowering quality.

Pick a task that is small, frequent and checkable

The best first tasks share three qualities. They are small enough to finish in one sitting, they come up often enough that a gain adds up, and you can tell quickly whether the output is right.

Summarising a long document into a page of key points.
Drafting a first version of a routine email or report.
Turning rough notes from a meeting into a list of actions.
Reformatting information from one layout into another.

Avoid tasks where a wrong answer is costly and hard to notice, such as figures that go straight to a client.

Write the request like a brief

An assistant does better when you treat it like a capable new colleague who knows nothing about your business. Say who the reader is, what the output is for, what to include and what to leave out, and show one good example if you have it. If the first answer misses, say what is wrong and ask again. The second round is usually the useful one.

Check everything that matters

These tools write fluently, and fluent is not the same as correct. They can state a wrong fact in a confident tone. Treat the output as a draft from someone fast and sometimes careless. Check names, numbers and claims against the source before they leave your desk, and keep a person responsible for every decision.

Mind what you share

Do not paste anything into a tool that your organisation has not approved for it. Customer details, contracts and unpublished figures need clear rules first. If the rules do not exist yet, ask for them.

Keep a small log of the task, the time it took and what you had to correct. After a few entries you will see where the assistant is dependable and where it keeps going wrong.

Do not wait to master the tool before you start. Skill comes from use, and the first try will feel slower than doing it yourself.

Measure, then widen

After two weeks, compare. How long did the task take before and after? Did the quality hold? Share what worked with your team, including the failures, and pick the next task together. Small, honest experiments teach more than a big announcement, and they leave you in control of how AI is used in your work.

Keep reading.

All articles