Instructor-led workshop for organisations ยท Managers and teams

Making Team Decisions with AI: who decides, and on what evidence

Each member of a management team now arrives at a meeting with an answer from their own AI assistant. This workshop teaches a team to use AI to prepare a decision faster, and then to decide together on the evidence: who owns the decision, who must be consulted, and what the team records.

Who should attend

Department heads and the teams that decide together: management teams, project boards and committees below board level. Run for a team that already works together.

Length

Half day or one day for one team, and a certification edition.

Delivery

In person, led by a facilitator, with decision practice on StepSim, the AIR APAC simulator.

Participants receive

Record of Completion for everyone. An optional, separate assessment can lead to an AIR APAC credential.

Status: designed, not yet delivered. Available to commission for your group. Case files and materials are completed for the first group that confirms.

Why this programme

The problem it solves

When each member of a team brings an answer from their own AI assistant, a meeting becomes a contest between confident drafts. The team agrees too quickly, or argues about wording, and no one checks the claim each draft rests on.

AI also makes it easy to act alone: past one's authority, before the people who must agree, and without saying that AI drafted the work. A team decision needs a named owner, the right people consulted, and a record of why.

What makes it different

Participants use AI to do the work, then check it

Most AI courses teach either how to use the tools or the risks of using them. This programme does both on the same task, in the same session.

Using AI

Tasks participants do with an AI assistant

  • Summarise the evidence pack before the meeting.
  • Draft two or three options, each with its costs and risks.
  • List the people affected and the objections each is likely to raise.
  • Draft the decision record and the message to the people affected.
Checking AI

What they learn to check before acting

  • Compare drafts from different people and different AI tools on the same evidence, not on how convincing they sound.
  • Find the claim that several drafts share, and check it once.
  • Bring disagreement into the open before the most senior or most confident view settles it.
  • Agree who owns the decision, who is consulted, and when it must go higher.
Outcomes

By the end, participants can

  1. Agree the real decision, and what is fixed, before anyone brings an AI answer.
  2. Compare options from different people and different AI tools on the same evidence, not on how convincing they sound.
  3. Bring disagreement into the open and test it, instead of following the most senior or the most confident view.
  4. State who owns the decision, who must be consulted, and when a decision must go higher.
  5. Record the decision, the reasons, the disagreement and what would make the team change course, and say where AI drafted the work.

What participants take back to work

  • A decision record the team uses for its next real decision.
  • An agreed rule for who owns which decisions, and when a decision goes higher.
  • A short checklist for meetings where people bring AI drafts.
  • One agreed change to how the team decides, with an owner and a date.
Agenda

How it runs

Half day or one day, for one team that already works together, not a mix of strangers. The team plays a StepSim Practice mission together its usual way, works through cases, and closes with an unseen mission of the same kind.

In the cases the team uses an AI assistant its organisation permits to prepare the decision, then runs the meeting on what it has checked.

The team leaves with one agreed change to how it decides when AI is in the work, with an owner and a date.

Each facilitated edition opens with a StepSim Practice mission played the participant’s usual way and closes with an unseen one of the same kind. The programme formats page describes the half day, the one day, the two days and the certification edition.

Formats

Half day or one day for one team, and a certification edition.

Cases

Composite team decisions written for the programme: a product launch that may need to slip, a budget cut shared across units, a change of a key supplier. Any AI tool in a case has "AI" in its name.

Method

All four domains, with most weight on Engage: Decision Rights and Escalation (E1), Stakeholder Alignment and Disclosure (E2), together with Problem Framing (S1), Assumption and Alternative Testing (T2) and Proportionate Commitment (P1).

After the programme

The half day ends with a written session report. The one day ends with a transfer file that goes back to the desk. The two days end with a transfer workshop, a report for the sponsor, and a manager review within ten working days.

See a sample exercise

Three confident drafts, one shared mistake

An example written for this page, in the style of the programme's cases. The organisations, people, figures and AI tools are fictional.

A management team must decide whether to delay a product launch by six weeks. Three members each bring a recommendation drafted with a different AI tool.

MemberAI toolRecommendationMain reason given
Head of sales"PitchAI"Launch on timeA competitor launches in four weeks
Head of product"PlanAI"Delay six weeksTwo open quality defects
Head of finance"LedgerAI"Launch on timeA competitor launches in four weeks; delay costs the quarter
What participants are expected to find
  • Two drafts rest on the same claim about the competitor's date. It came from one news summary and no one checked it.
  • The quality defects appear in only one draft, so the meeting treats them as one person's view.
  • No draft says who owns the decision. The head of sales and the head of product each assume it is theirs.
  • Customer service, who would handle the defects, was not consulted.

What the facilitator observes

In the StepSim Practice missions, the facilitator looks at how people decide, not at whether they reached a particular answer. For an exercise like this one:

  • Whether the team checked the claim two drafts shared before deciding.
  • Whether the decision owner was named before the discussion, not after.
  • Whether disagreement was recorded with its reason.
  • How these change between the first Practice mission and the last one.

This is practice evidence from the session. It shows what changed in how the group checked AI work during the programme; it is not proof of better results at work, and Practice never counts toward certification.

Instructors

Who teaches the programme

Judgment with AI

Lead facilitator

For a first delivery, the lead facilitator is Indhran Seghar, director of AIR APAC Pte. Ltd. and the author of AIRS-J/1, the standard this programme is built on. Later deliveries may be led by a training provider's facilitator, named in the proposal.

Certified facilitators

What “certified” means here

AIR APAC certifies facilitators under section 20.6 of AIRS-J/1. No facilitator holds that certification yet, because the first facilitator cohort has not run. If your organisation needs a trainer accredited under a national training framework, say so in your request; we will tell you plainly whether it can be met.

Credentials

What participants receive, and what is optional

WhatWhat it isIn this programme
Record of CompletionA record, issued by AIR APAC, of what the person attended and practised. It carries no score or level.Included for everyone who completes.
AIRS-J/1 certification, Professional trackAIR APAC’s own credential for judgment in AI-enabled work, under a published standard. Verified free by code on a public register, valid for two years. No regulator, university or professional body recognises it, and no certifying sitting has run yet.Optional. The certification edition adds a separate half-day assessment sitting within 30 days of the programme. Attending the programme does not count toward the result.
A qualification from an outside professional bodyA qualification, or continuing professional development hours, from a professional body in your field.Not offered. It would need an arrangement with that body, which does not exist today.

How AIRS-J/1 certification works.

Questions from buyers

Before you request a proposal

What are StepSim, STEP and AIRS-J/1?

StepSim is a decision simulator, an AIR APAC initiative built and operated by Exeter Labs. Participants play short workplace missions in which AI tools give advice, and they decide what to do. STEP is the method every AIR Academy programme teaches: four questions a person asks before acting on an AI answer (Structure, Test, Engage, Perform). AIRS-J/1 is the AIR APAC standard for judgment in AI-enabled work, which defines those four areas.

Can it be fitted to our sector, our rules and our language?

The cases are written for the programme and can be set in your sector and your country’s rules; say which in your request. Cases about your own organisation are used only with its written consent. Programmes are delivered in English today; if your group needs another language, say so, and we will tell you plainly whether it can be arranged.

In person or online, and how many people?

The facilitated editions are led in the room. The half day is built for 15 to 30 people; group size for the longer editions is agreed in the proposal. Whether a live online edition is possible is also agreed in the proposal.

How is the price set, and how soon can it run?

The price is quoted per group in the proposal. It depends on the edition, the number of participants, the location, and whether the certification edition is included. No price is published. The programme is designed but has not yet been delivered: case files and materials are completed for the first group that confirms, so the earliest date is agreed in the proposal.

Will our confidential information go into AI tools?

No. The exercises use fictional case files. Where participants work on their own tasks, they use only an AI tool their organisation permits for that information, remove identifying and confidential details first, or use a fictional version instead. Nothing they show in the session is kept or recorded, and it is never used for certification. Your organisation receives these conditions before the programme starts.

How will we know it worked?

Each edition opens with a Practice mission played the usual way and closes with an unseen one of the same kind, so the group can see what changed in how it checked AI work. The half day ends with a written session report. The one day ends with a transfer file that goes back to the desk. The two days end with a transfer workshop, a report for the sponsor, and a manager review within ten working days. This is evidence from the programme, not proof of better results at work.

Does it lead to a recognised professional qualification?

Not on its own. Everyone who completes receives a Record of Completion. Participants who take a certification edition can earn AIRS-J/1 certification, issued by AIR APAC; no regulator, university or professional body recognises it. See the credentials table.

For training providers

Bring this programme to your clients

AIR Academy programmes are delivered through training providers, who own the client relationship and invoice the client. For a first delivery AIR APAC provides the programme design, the case files, StepSim Practice missions and the lead facilitator. The fit to your client, the subject co-facilitator and the commercial terms are agreed directly; no terms are published. Facilitators who want to run StepSim sessions in their own programmes can join the facilitator programme.