What AI can take off your team's desk this week

By Stacy Maynard

When a team asks me where to start with AI, I ask them a different question first: what's the job everybody puts off?

Usually it's the meeting notes nobody writes up, the newsletter that goes out late every month, the proposal rebuilt from scratch each time, the same twelve questions answered by email over and over. Those are the jobs where AI earns its keep, and they're where I'd start.

The jobs it does well right now

Meeting notes into actions. Record the meeting (with everyone's agreement), let a transcription tool capture it, then ask an AI assistant to pull out decisions, owners and deadlines. Someone still reads it before it goes out, but the hour of writing becomes five minutes of checking.

First drafts of anything repeated. Proposals, quotes, grant sections, job postings, policy updates. Give the assistant two or three of your best past examples and it will draft the next one in your structure and your voice. The person who knows the work then edits it, which is a far better use of their time than staring at a blank page.

Answering the same questions. Most organizations have a handful of documents that answer most of what people ask: the member handbook, the FAQ, the policies. Load those into a shared AI project and staff can get an answer drawn from your own material, with the source, instead of searching three drives.

The newsletter and the social posts. One update from the director, one event, one success story. AI can turn that into a newsletter, a few social posts and a short email for the board. More on that in our post about building a content system.

Tidying the inbox. Sorting, summarising long threads, drafting replies to routine requests. Small on its own, but it adds up across a team.

Start with one tool, set up properly

The biggest difference I see between teams that get value from AI and teams that don't is setup. Half the staff using different free tools, each starting from nothing every time, gets you very little.

Pick one approved tool. Create a shared project for the team and load it with the things that make your organization yours: your style guide, your common documents, a few great examples of past work, and plain instructions about tone. Now every person starts from the same place, and the answers sound like someone who works there.

Write the rules down

A one page policy does more for safe use than any training video. It should say which tools are approved, what can be pasted in and what must never be (client personal information, health details, anything confidential), and who to ask when unsure. Keep it short enough that people actually read it.

Two habits belong on that page too. A person checks everything before it goes to a client, member or the public. And if AI wrote most of something that's published under the organization's name, someone who knows the subject has read every line.

Measure it, roughly

You don't need a dashboard. Before you start, ask the team how long the chosen job usually takes. A month later, ask again. If the newsletter went from a day to two hours, you have your answer, and a good argument for the next job on the list.

Where to go from here

If you want your team doing this on their own work rather than reading about it, our Claude at Work session runs for 90 minutes and ends with a working project and three real tasks finished. For teams that want the whole setup built and handed over, that's what an AI Unit is for.

Not sure which you need? Take the free AI Readiness Scorecard. Ten questions, about four minutes, and you'll get the first three things we'd fix.

Questions people ask about using AI at work

What tasks can AI do for a small team?

Meeting notes turned into action lists, first drafts of documents you write again and again, answers drawn from your own handbook, newsletter and social drafts, and inbox sorting. A person still checks everything before it goes out.

Is it safe to use AI with company information?

It can be, with an approved tool and clear rules. Never paste in client personal information, health details or anything confidential, and put those rules on one page so everyone knows them.

Which AI tool should a small team use?

The one your organization approves and can support. Set it up properly as a shared project with your documents and examples, because one tool set up well does more than several free accounts.

How do you measure whether AI is saving time?

Ask how long a task takes before you start, then ask again a month later. If the newsletter went from a day to two hours, you have your answer.

Do we need an AI strategy before we start?

You can start smaller. One job the team dislikes, one approved tool and a one page policy will teach you more than a planning document, and the strategy gets easier once you've seen it work.

Read next: Your Google Business Profile is the new front door and One conversation, a month of content.