Get your data ready for AI
Find out whether AI has the right business information, and whether that information is safe to use.
Why AI is only as useful as your information
Lesson 1 of 3 · Why AI needs your own information
AI can produce answers without your business information. It needs relevant, reliable business information to make those answers useful for your specific work.
A general AI assistant knows a lot about the world. It does not know your prices, your contracts, your customers or what happened last month. When you ask it about your business, it fills the gaps with what sounds likely. The answer looks complete either way.
The question is not "is our AI good?" It is "what information is the AI using for this answer, and is it the right information?"
Lesson 2 of 3 · Seven checks before AI relies on your information
Data readiness means knowing what information exists, who owns it, and whether it is safe and suitable for the job you want AI to do. Run seven checks, for one use of AI at a time:
- Source and owner. Do you know where this information comes from, and which person is responsible for it?
- Quality. Is it complete and correct, without large gaps, duplicates or old figures?
- Access and permission. May it be given to an AI tool under your own rules, your contracts and the law?
- Freshness. Is it updated often enough for this use?
- Meaning. Do the AI and your people read each item the same way? (A "customer" or a "sale" can mean different things in two departments.)
- Tracing an answer. Can a person see which records an AI answer came from?
- Testing. Have you tested the AI on questions where you already know the right answer?
Check one use at a time. The same records can be ready for one use and not ready for another.
Lesson 3 of 3 · A confident answer from incomplete records
Suppose an AI tool recommends a supplier. It relies on current prices, delivery history and the conditions in each contract. Now suppose delivery records are kept in one region only.
The ranking will favour suppliers from that region, and the recommendation will still look confident. Nobody will see the problem by reading the answer.
The fix is not a better AI tool. It is to know which records the ranking used, who owns them, and what is missing. Then test: ask the AI questions where you already know the right answer, and see where it goes wrong.
If you cannot say which records an AI answer came from, you cannot know when it is wrong.
Check one use of AI against seven questions
Choose one use of AI your team wants, and the one source of information it depends on most. Answer honestly: "Not sure" is a useful answer.
When the gaps belong to other people
If your checklist shows gaps, you have found the next problem. The information exists, but it is often in several systems, some records disagree, and nobody has decided what to fix first, how much work it is, or how AI may use it safely.
Data readiness briefing
Your team brings one proposed use of AI. We go through the seven checks with the people who own each source. You leave with a list of gaps in order of priority, with an owner for each. AIR APAC does not take copies of, clean, move or store your data; if technical data work is needed, it is scoped separately.
The next free programme
You have identified the information AI needs. Who will recognise when its recommendation is still wrong?
AIR Academy free programmes
AIR Academy is the training arm of AIR APAC, the Center for AI Readiness in Asia Pacific, a social enterprise company based in Singapore. This programme is free, needs no account and stays free. Your answers in the exercise stay in your browser: nothing is sent to AIR APAC.
The three free programmes: People: From Answers to Decisions · Process: Who does what when AI joins the work? · Data: Get your data ready for AI