Do you actually need AI for this?
Sometimes the honest answer is no, and that's a good result. I started Nortevant to bring AI in only where it's appropriate, so this guide gives you a quick way to test one task yourself.
You're in good company
In the government's 2026 survey, 65% of businesses not using or planning to use AI cited a lack of a clear business case or proven return. Waiting for a good reason is a perfectly sensible position.
Pick one task
Don't start with "AI". Start with one job your team does regularly, ideally one that makes people sigh. The more specific it is, the better: "typing delivery notes into the system", not "admin".
Ask five questions
- Does it happen often? A job done twice a year won't repay the effort of setting anything up.
- Is the work mostly the same each time? If every case is different, the thinking is probably the job.
- Would a mistake be easy to spot and fix? If not, you'll spend the time you saved on checking.
- Can it be done without sensitive information? If it needs client or staff details, the tool and its settings matter a lot more.
- Would your customers be comfortable if they knew? If not, keep it with a person.
Four or five yeses is worth a try. Two or three means look closer. Fewer than that, and I'd leave it alone. It's a rule of thumb from me, not a scientific score, but it should get you to an honest answer.
Try the simpler fix first
Some problems that look like AI problems aren't. If people retype the same details into two systems, the answer might be connecting those systems, or a template, or stopping the task altogether. It's worth asking why the job exists before asking how to speed it up.
Your existing software may also already do what you need, so check that before paying for anything new.
If the answer is no
Then you've saved money and avoided a distraction. Write down why, and look again in six months, because the tools change quickly. A clear "not yet" is a better place to be than a subscription nobody uses.
If the answer is yes
Try it small: one person, one task, for two weeks. Keep it away from client data while you're testing, and note how long the job took before you began. If it doesn't clearly help, stop.
