Part 5: Giving AI your knowledge

What is a knowledge base, and does your business need one first?

Written lesson · about 3 minutes to read

The short answer

A knowledge base is an organised, current collection of your business's answers and documents that people and AI tools can search. You do not need a perfect one first: start by writing down the twenty questions asked most often, each with its answer and a named owner.

Three things to remember

  1. 01If the answers live in a few people's heads, an AI tool has nothing to read.
  2. 02Start with the twenty questions your team asks most, each written down with an answer and an owner.
  3. 03Capture what experienced people know before they retire or leave, starting with the one closest to going.

What a knowledge base is

A knowledge base is your business's answers, kept in one place, organised and up to date, where people can search for them. It can be a shared folder with clear file names, a set of shared pages, or a collection inside an AI tool. The tool matters less than the habit: each answer is written down, has an owner, and gets changed when the facts change.

The test for what belongs in it is simple. If a new starter would otherwise have to interrupt a colleague to find out, it goes in: the price book, how a warranty claim is handled, which part fits which unit, who to call when a supplier is late.

Why AI needs one first

An AI tool that answers from your documents can only read what has been written. If the way you price a job, or the reason the old boiler needs a particular fix, lives in two people's heads, the tool has nothing to find. A badly set up tool may then give a general answer that fits some other business.

It depends on where you are today. If most answers already sit in documents, you need tidying, not a new project. If most sit in people, writing comes first. That work pays off even if you never buy an AI tool, because every new starter benefits too.

Start with twenty questions

Do not try to document the whole business. List the twenty questions your team asks most often, the ones that make someone stop work to ask a colleague. Write each answer in plain words, note the source document if there is one, and put a name against it as the person who keeps it right. That short list is already a working knowledge base. Add to it as new questions appear.

Capture what experienced people know

Some of the best answers sit with someone who has done the job for years and never wrote it down. Getting them out has to be easy or it will not happen. A forty-second voice note on a phone after a job works better than asking someone to sit down and write a manual.

Ask four things: what do you check before you start, what does it look or sound like when it is going wrong, what do you do that the instructions leave out, and what did you learn the hard way? Let the person read the written version once, then file it where everyone can find it.

Where to start this week

Ask three or four people to jot down every question they had to look up or ask about this week. Pool the lists, pick the twenty that come up most, and write the answers. Name an owner for each and put a review date on it.

Then pick your most experienced person, the one closest to retiring or leaving, and record the first voice note this month. What they know cannot be recovered afterwards, and the twenty written answers give an AI tool something real to read.

Next step

Where does this fit your business?

Six quick questions about your business, then your AI plan: what to start with, what a person still checks, and the lessons that go with it. Or bring it straight to a free thirty-minute call.