Key points
- Before starting with AI, a business can check that it is ready. It need not be perfect, but four things should be good enough: information, people, process, and someone in charge.
- First, information. Find where it is kept: in one place, or spread over email inboxes, shared folders and memories. Check it is current, with no old versions clashing with new, because an AI tool fed stale documents gives stale answers, fast and politely.
- If that information holds personal details of customers or employees, get a qualified adviser's view on your duties before passing any to a supplier outside the business.
- Next, people. Ask whoever does the task now, without pointing fingers, whether AI tools already play a part in their work, and how. Many staff do, often on private accounts. You may find good ideas, or data going where it should not, so find out first.
- The people who do the work know best whether a new tool suits how the job really works, so bring them in early.
- Then, process. If no one can explain the task's current steps, a consultant will charge you to work them out. That stage, often called discovery, can be worthwhile, but choose it rather than stumble into it. Sketching the steps yourself, even roughly, makes later work cheaper.
- Finally, an owner: one person for the project and, later, the system it creates. In a micro business, probably you. In a bigger organisation, a manager with power to reshape the work, not just an IT person, because unowned projects wander and unowned systems run down.
- And whoever approves the budget should expect some results to be wrong, and plan time for staff to check them. Get these four things good enough, and you are ready to start.
Transcript
Before starting with AI, a business can check that it is ready. It need not be perfect, but four things should be good enough: information, people, process, and someone in charge. First, information. Find where it is kept: in one place, or spread over email inboxes, shared folders and memories.
Check it is current, with no old versions clashing with new, because an AI tool fed stale documents gives stale answers, fast and politely. If that information holds personal details of customers or employees, get a qualified adviser's view on your duties before passing any to a supplier outside the business.
Next, people. Ask whoever does the task now, without pointing fingers, whether AI tools already play a part in their work, and how. Many staff do, often on private accounts. You may find good ideas, or data going where it should not, so find out first.
The people who do the work know best whether a new tool suits how the job really works, so bring them in early. Then, process. If no one can explain the task's current steps, a consultant will charge you to work them out. That stage, often called discovery, can be worthwhile, but choose it rather than stumble into it.
Sketching the steps yourself, even roughly, makes later work cheaper. Finally, an owner: one person for the project and, later, the system it creates. In a micro business, probably you. In a bigger organisation, a manager with power to reshape the work, not just an IT person, because unowned projects wander and unowned systems run down.
And whoever approves the budget should expect some results to be wrong, and plan time for staff to check them. Get these four things good enough, and you are ready to start.