Trust
What evidence would make me rely on this?
Learn to distinguish one good result from evidence that a service is dependable, then choose what may run while important evidence is still missing.
Required: your AI Operating Charter and either your own workflow or the fictional starter case. AI/GPT is not required for the core simulation.
Inspect a good draft, name what it proves and does not prove, choose a provisional operating mode and challenge it with an exception.
A provisional operating decision, explicit evidence gap and exception boundary.
This session uses supplied records and decision simulations. Do not treat the exercise as proof of a live production control.
What you will leave with
A provisional operating decision and evidence gap. Add this to your AI Operating Charter.
Use your own workflow or the fictional quotation case. No coding or live customer action is needed.
Start with a decision · 10 minutes
Show one good quotation draft. Ask: would you leave this service answering customers tonight? Require a reason and one missing piece of evidence.
Bite A · Name the useful service
Question: What benefit does this draft actually demonstrate?
Input: S0–S4, especially corrected Reply B. Open the case cards.
Try · 12 minutes: List useful preparation already visible, who benefits and what remains unproven. Distinguish a draft result from repeated service performance.
Compare after you have decided
The draft preserves current price and asks for missing details. It proves neither consistent quality nor authority to send automatically.
Keep: Purpose, users and useful outcome.
Bite B · Choose a provisional operating decision
Question: What may run while evidence is incomplete?
Input: Reply B and shared action modes. Open the case cards.
Try · 12 minutes: Choose Prepare or another defensible mode with conditions. Name the most important evidence gap and the next observation needed.
Compare after you have decided
Prepare keeps useful drafting available with human review. “Not enough evidence” should identify the missing evidence and a next step.
Keep: Current mode, condition and next evidence.
Bite C · Challenge confidence with an exception
Question: Does a good normal case cover a discount exception?
Input: Q-106 asks for 15% off; S2. Open the case cards.
Try · 12 minutes: Identify what may continue and who decides the discount. Revise the provisional decision if you were assuming every enquiry was normal.
Compare after you have decided
Daniel decides above 10%; the assistant may prepare. A successful ordinary draft does not establish exception handling.
Keep: Scope boundary and exception owner.
Put it together · 20 minutes
A peer can name the service, its benefit, current permitted mode and evidence still needed before greater dependence. Revise the weakest item and have a partner use it. Keep the result and one limitation.
Close your question · 10 minutes
Return to What evidence would make me rely on this? and your own opening question. Answered, partly answered or still open? Show the evidence. Save the updated AI Operating Charter. A reasoned decision to keep a step human-led is useful work.
Common trap
Do not use a polished demo or provider reputation as proof of this service’s reliability.
Must I already operate AI? No. The supplied records let you make the same operating decisions without a deployed system.
Where this appears in real work
Alamak: a strong generated story is one result; a dependable publication process needs additional evidence. The class uses the fictional packet; selected project demonstrations and technical walkthroughs can add detail later.