AIR APAC · Free self-check · People, Process, Data

Is your organisation ready to work with AI?

Choose one piece of work and answer 15 short questions about the people, the process and the data behind it. You get your next three decisions and who to talk to.

Start the check

Before you start

Choose one piece of work

Pick one piece of work where your team uses AI, is testing it, or plans to. Every question is about that work and the team that does it.

The work

The AI tool (optional)

15 questions, about six minutes. No score.
Answer for what is true today, not what is planned. If AI is not used yet, the first answers are often the honest ones.
Your answers stay in this browser unless you choose to send them to us. Reloading the page clears them.
A preliminary self-check, not an assessment.

Your result

Your next decisions

Chosen from your answers that describe little or no practice today, the ones with the most risk first.

Questions to take back

You answered "Not sure". Find out before you decide anything else.

Take this to your next management meeting

Get a one-page meeting brief by email. It is written so you can forward it unchanged:

your three decisions, set out as a meeting agenda;
what to ask Finance, Operations, Technology and People;
an owner-and-date worksheet, and an order for the next 30 days.

With it, a ready-to-run team exercise: five past cases from your work, two with a planted AI error. Thirty minutes.

Work email

Add your name and organisation (optional)

Name

Organisation

Your role

People who do this work

Are you planning to change this work in the next six months? (optional)

Yes No Not sure

What would help next? (optional)

A session for my team A review of this work Information for later

You may contact me about a session for my team. (optional) Include my fifteen answers. (optional; without this we receive only the work, the AI tool and the question numbers behind your result)

We use your email to send the brief and, only if you tick the box above, to contact you about a session. AIR APAC does not share it with anyone, including PIKOM or event organisers. We keep the request in our enquiry records (Notion, a service based in the United States) for twelve months, then delete it. Reply "stop" to any email from us to withdraw. Privacy

Discuss with your colleagues

Ask one or two colleagues who know this work to answer separately, then put your answers side by side. Where you differ is often where to start: operations may think a person approves every change while IT knows the tool can act alone.

All your answers

Learn more

PeopleTraining for decisions with AI ProcessWho does what when AI joins the work DataSeven checks before AI relies on your information Free programmeDeciding well when AI gives you the answer

A preliminary self-check, not an assessment. There is no score and no comparison with other organisations. This self-check is separate from the AI Readiness Index and does not change any Index result. AIRS-J/1 certifies a person's judgment. It does not certify an organisation's readiness.

Your browser is not running JavaScript, so the check cannot show a result. You can print these questions and discuss them with your team, for one piece of work.

People: AI gives the answer. Your people make the decision.

How does the team know which AI tools it may use for this work?

Nobody has told them.
A manager or colleague gives informal advice.
A written list says which tools are approved, and for what.
Someone has checked that the team uses only the tools on the list.

What must the team check before it uses an AI answer in this work?

Nothing is agreed. Each person decides.
Checking is encouraged, but nobody has said what to check.
A written instruction says what to check.
A recent review of finished work showed that people did the check.

Who is responsible for decisions in this work that use AI?

Nobody is named. People say the system decided.
Whoever used the tool, by default. Nothing is written down.
A named job role is responsible, and this is written down.
Records of important decisions show who approved them and why.

How have people in the team practised checking AI on this work?

They have not. They learned by using the tool.
General AI talks or online courses.
Practice on real examples from this work.
Practice on examples with AI errors in them, with feedback on what they missed.

If someone in the team finds a serious AI error, what can they do?

There is no agreed way to report it.
They can tell their manager. What happens next is unclear.
They know who to report to, and may stop using AI on that case.
Reported errors are reviewed, and the team hears what was changed.

Process: Who does what, where a person checks, who approves.

Is this work written down, step by step?

No. Only the people who do it know the steps.
Some steps are written down. Handovers and unusual cases are missing.
The steps, handovers and unusual cases are written down.
The team checked the document against recent real cases and corrected it.

Has it been decided, step by step, what AI may do and where a person must check?

No. Each person decides as they go.
It has been discussed, but nothing is written down.
It is written down, including where a person checks and who approves.
The owner of the work approved it, and it is reviewed when the AI tool changes.

What stops the AI from taking an action it should not take in this work?

Nobody has decided which actions the AI may take.
Limits have been discussed, but nothing is written down.
Allowed actions, and those that need a person's approval, are written down.
The tool itself blocks actions outside those limits, and records what it does.

How would an AI error in this work be found?

It might not be found until a customer or an auditor sees it.
A person might notice while doing the work. There is no required check.
A required check catches errors before the output is used.
We tested this with deliberately wrong answers, and errors are logged.

Has a measure been agreed to show whether AI makes this work better?

No measure has been agreed.
We have estimates of the benefit, but no starting number.
One measure is agreed, and the number before AI was recorded.
We compare before and after, including errors and rework, not only time saved.

Data: No data, no AI.

Do you know which records, documents and systems the AI uses for this work?

The sources have not been listed.
People know the main sources, but there is no agreed list.
There is an agreed list, with a person responsible for each source.
For any AI answer, a person can see which records it came from.

Has the information the AI uses been checked for errors?

It has never been checked.
People know of problems, but no sample has been checked.
A sample was checked for the errors that matter to this work.
Samples are checked again after changes, and problems go to the person responsible.

How does the team know which information may be given to an AI tool?

Nobody has checked whether it is allowed.
People decide case by case. Nothing is written down.
Written rules, confirmed by legal or data protection, say what may be used.
Access is set so that only approved people and tools can reach this information.

Do you know where the information goes when AI uses it, including AI features your software vendors have switched on?

We have not checked.
We know the tools we chose, but not the AI features inside our existing systems.
We have a list of every AI tool and feature that reads this information.
For each one, we know where the information is stored, for how long, and in which country.

Has the AI been tested on past cases from this work where you know the right answer?

It has not been tested on past cases.
People tried a few examples. The results were not written down.
A set of past cases, including difficult ones, was tested and the results recorded.
The test is run again whenever the tool or the information changes.