AI readiness · Pillar 1 of 3

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

AI can help people work faster. Speed is not enough. Your staff need to know when to use AI, how to check its work, when to ask for help, and which decisions remain their responsibility. AIR Academy, AIR APAC's learning division, trains managers and teams to do this.

What people need when AI is part of the work

Three kinds of skill. Most organisations need all three, for different people.

Check the answer before acting on it

Ask which question the AI answered, test the figures it rests on, bring in the people who must agree, and record the decision with an owner and a date.

From Answers to Decisions

The programme

From Answers to Decisions: practical AI training for managers who approve decisions

Each edition of this programme starts with a simulated decision made your usual way and ends with a new case of the same kind, so the group sees how its checking changed. The cases are written for your sector. It runs as a half day, one day or two days.

  • Half day: a written session report.
  • One day: each participant leaves with a written change to how their team checks AI work, with an owner and a date.
  • Two days: a workshop on applying the method at work, a report for the sponsor, and a manager review within ten working days.
See the programme Host the free one-hour session
Illustrative scenario

Same AI recommendation, two decisions.

Meridian Supplies is a fictional distributor with 38 staff. Its AI assistant, KessidAI, recommends a USD 12,000 a year reporting tool, saying it will save about 120 staff hours a month. Two managers at Meridian receive the same recommendation. Both have the authority to approve it.

Manager A

  • Approves the subscription this week. The capacity value is more than three times the cost.
  • Does not ask what the 120 released hours would be used for.
  • Does not ask who will check the automated reports before the owner reads them.
  • Sets no condition for review and no date to check.

Manager B

  • Asks the vendor to run Meridian's own weekly report through the tool, and counts the staff hours that takes.
  • Asks finance how a released hour turns into money at Meridian.
  • Asks the three people who read the Monday report what they need from it.
  • Decides on a pilot on two reports, with an owner, a review date and the result that would justify the full subscription. Records that KessidAI drafted the original case.

Both managers saw the same numbers. The second wrote a decision another person could check. That is judgment: how well a person decides once AI has given the answer. AIR Academy programmes give participants practice in it. A separate assessment evaluates it against AIRS-J/1.

Illustrative scenario built from a fictional case. Not participant data, and not evidence of programme results.

Inside StepSim

See how a decision is recorded in StepSim

These are real screens from one mission in StepSim, built by Exeter Labs: a facilities tender where an AI tool ranked the suppliers. The participant reads the AI advice, checks the evidence, decides, and then sees what happened three months later.

The desk: an AI assistant recommends awarding the facilities contract for all five regions to the supplier an AI scoring tool ranked first, and says it is 86% sure. Each piece of its advice can be used, pushed back on or set aside. Files and colleagues cost review hours.
The desk: an AI assistant recommends awarding the facilities contract for all five regions to the supplier an AI scoring tool ranked first, and says it is 86% sure. Each piece of its advice can be used, pushed back on or set aside. Files and colleagues cost review hours.
Checking the evidence: the category manager's note shows site work the tender never priced, such as hospital cleaning and lab waste. The participant can now use this evidence to limit the award.
Checking the evidence: the category manager's note shows site work the tender never priced, such as hospital cleaning and lab waste. The participant can now use this evidence to limit the award.
The decision: award as the AI proposed, award with limits, hold and keep the current contracts, run the tender again, hand the decision up with a recommendation, or send nothing and let the deadline pass.
The decision: award as the AI proposed, award with limits, hold and keep the current contracts, run the tender again, hand the decision up with a recommendation, or send nothing and let the deadline pass.
Three months later: what happened after the award, the actions colleagues took and the money against plan. A separate screen then gives each participant a written record of how they decided.
Three months later: what happened after the award, the actions colleagues took and the money against plan. A separate screen then gives each participant a written record of how they decided.
See the programme that uses this mission
Standard and certification

A public Standard for judgment with AI

AIRS-J/1, the Standard for Judgment in AI-Enabled Work, publishes nine skills in four areas, with scoring rules anyone can read. A separate assessment on StepSim measures each person against them and can lead to certification.

AIRS-J/1 certifies a person's judgment. It does not certify an organisation's readiness.

Read the Standard · How certification works

Where to start

Training for your team

Plan training for your managers

Tell us the team, its size, the decisions it makes with AI, and when you want to start. We will reply with a proposal.

Explore training programmes Request a proposal

The three pillars of AI readiness

People, Process and Data, in that order. An organisation needs all three to use AI well.