Indhran SegharJudgment when working with AI

Applied work

What judgment looks like in the work.

A useful AI answer still leaves a decision to make. These examples show the evidence to check, the conditions to record and the person who remains responsible.

I use public material from four ventures to explain the decisions behind the work. The organisations and public material are real; the scenarios below are demonstrations or teaching examples. They are not client case studies or measured results.

Tradezur · Public hypothetical samples

Is this cross-border opportunity worth pursuing, and is the counterpart as presented?

Tradezur examines cross-border commercial fit, counterpart identity and demand evidence. Its public buyer sample returns “clarify” because demand and timing are unestablished. A separate distributor demonstration holds exclusivity pending confirmation of the importing entity and the import-for-resale arrangement. These are hypothetical examples, not forecasts or client results.

How the decision develops

  1. Test demand evidence and timing before the counterpart conversation goes further.
  2. Confirm the importing entity and the import-for-resale arrangement before exclusivity is discussed.
  3. Name the next action, owner and date for each open question.

Questions to take into a meeting

  • What evidence shows demand, beyond stated interest?
  • Who is the importing entity, and is this import for resale?
  • What would move this from clarify to pursue, and by when?

What I take from it

A reasoned recommendation is only as good as the evidence under it. Where demand, timing or identity is unestablished, clarify is the honest output. Named actions, owners and dates make it useful; legal compliance is not implied.

Explore the public work at Tradezur. Source checked 6 November 2026; public examples may change.

Robometic · Public simulated evidence pack

Can this cooling equipment be taken offline under present operating conditions?

Robometic develops engineering intelligence for data centre cooling. Its public simulated example considers taking one pump offline while another supports the load. Start and stop checks passed, but full-load capability is unverified; headers and controls are shared, and the electrical path was not assessed. The pack does not authorise isolation. The authorised site team retains that decision.

How the decision develops

  1. Read the evidence pack against the present operating conditions.
  2. Separate what was tested from what was not: full load, shared headers and control, electrical path.
  3. Return the decision to the authorised site team with the basis stated.

Questions to take into a meeting

  • Which parts of the present condition are evidenced, and which are assumed?
  • What does a start and stop check leave unverified?
  • Who holds authority for isolation, and what would they need to see?

What I take from it

An evidence pack can describe a decision without authorising it. Keep tested and untested apart, and leave the decision with those who hold authority and site knowledge.

Explore the public work at Robometic. Source checked 6 November 2026; public examples may change.

Tourism Room · Public programme demonstration

Is the programme producing additional local benefit, or recording purchases that would have happened anyway?

Tourism Room presents a 90-day district programme connecting local businesses, hotels, agents and operators through existing booking channels. Its demonstration reviews purchases, refunds and operator effort so a sponsor can consider whether to continue, change or stop. The demonstration does not establish additional local spending or a delivered programme result.

How the decision develops

  1. Define what counts as a purchase and what counts as additional local benefit.
  2. Read purchases, refunds and operator effort together, since each changes the picture.
  3. Put continue, change or stop to the sponsor with the basis stated.

Questions to take into a meeting

  • Which purchases would have happened without the programme?
  • What did the programme cost in operator effort, and who carried it?
  • What evidence would change the sponsor's decision?

What I take from it

Purchases are countable; additional local benefit is not the same thing. Keeping the two apart, and stating the limit of attribution, is what makes a programme review honest.

Explore the public work at Tourism Room. Source checked 6 November 2026; public examples may change.

AIR APAC · Published teaching and fictional exercises

What should a manager check before owning this decision?

AIR APAC, the institution I founded, brings People, Process and Data together through training, research and standards. In a fictional supplier or reporting-tool decision, a manager checks assumptions, follows approvals and uses current source evidence. The free worksheet on this site lets you practise those questions; institutional learning belongs at AIR APAC.

How the decision develops

  1. Check the assumptions in the material against current source evidence.
  2. Follow the workflow approvals rather than the shortest route to a decision.
  3. Name the decision owner and record the basis for the decision.

Questions to take into a meeting

  • Which assumptions are mine to check, and which are already evidenced?
  • Where does the approval route sit, and who owns each step?
  • What is the current source, and how would I know it changed?

What I take from it

A useful exercise ends with a person who can explain the evidence, conditions and approval. Practising these questions is a starting point for learning.

Explore the public work at AIR APAC. Source checked 6 November 2026; public examples may change.

Apply the questions to your own decision.

Start with the free decision worksheet. For a team session, choose a topic and compare formats and fees.

Contact me