Slide 1 — Cover and outcomes
Introduce the designed scenario and the three outputs; clarify that this is a teaching sample, not a real client record.
Learner action: identify which output is most relevant to their work. Objective L1–L3. Proposed teaching time: 1 minute.
Slide 2 — The problem
Invite a quick observation; do not claim this was a measured employee mistake. The example draft was written for this lesson, not generated by a model.
Keep the explanations back — ask for one detail that would need checking before you say where it came from. All three borrowed details are discussed on slide 6.
Learner action: name one detail that needs checking. Objective L2. Proposed time: 2 minutes.
Slide 3 — The method
Missing responsibility and permission are business issues, not facts for an AI to invent. Advance the three steps manually; each step shows the concrete mini-artifact the learner ends up with.
Learner action: point out where a human must decide. Objective L1–L3. Proposed time: 3 minutes.
Slide 4 — The prompt
Explain that the source itself must be suitable for the organization's approved tools. The four layers are a contract: the boundaries layer is what prevents borrowed owners and invented quantities.
Open the full prompt and copy it, or select the text if clipboard access is blocked. No AI service is called from this page.
Learner action: identify the instruction that prevents guessed task owners. Objective L1–L2. Proposed time: 4 minutes.
Slide 5 — Source to action table
Reveal one step at a time and ask where the date comes from. The 6 November event date in the context strip belongs to the briefing, not to any task deadline.
In V2 the “Why this field” panel answers that question in writing, so a learner working alone still sees the reasoning. Reveal one step at a time; the panel is a teaching aid, not a claim about a real AI output.
Learner action: after each step, name one filled field and one field that stays TBD. Objective L1–L2. Proposed teaching time: 5 minutes.
Slide 6 — Check the draft
Alex and 4 November are valid for another task, which is precisely why a careless draft can look plausible. The visitor count is different — it has no source at all.
In V2 the learner selects each span themselves; the trainer no longer presses a single Reveal. Ask why each selected span is unsupported for this task. No score is given.
Learner action: select each span and state why it is unsupported for this task. Objective L2. Proposed time: 3 minutes.
Slide 7 — Independent practice
Let learners attempt before revealing. The reference is not authorization to buy office supplies.
In V2 the learner types directly on the page; the trainer no longer controls a single reveal. Entries are held in memory only — they survive moving between slides but are cleared by a refresh or by closing the page. Nothing is saved or sent, and there is no grading.
Learner action: draft the summary, the three action rows and the clarification questions, then compare with the reference. Objective L1–L3. Proposed time: 4 minutes.
Slide 8 — Take it into work
Ask the learner to choose one low-risk task to practice using approved tools. The template is a local plain-text file — nothing is collected or transmitted.
Learner action: state one application and its confirmation boundary. Objective L1–L3. Proposed time: 2 minutes.
Scope and version
Self-study sample V2 — exactly eight pages, proposed 20–30 minutes (page timings total 24 minutes, planning estimates only). This is one module of a 180-minute workshop proposal, not the complete workshop.
The demonstration on slide 5 is pre-authored; no live model call is made or implied. V2 adds self-study support on slides 5, 6 and 7 only: written “why this field” explanations, per-span error inspection, and editable practice with hints and a reference answer. Slides 1–4 and 8 are unchanged from V1. Practice entries are in-memory only and are never transmitted.