Personas
All personas are fictional. They are design tools synthesized from the secondary research and the needs analysis — not real people, not survey respondents. Quotes are invented to summarize each persona’s situation.
Last reviewed: 2026-10-25
Workplace track
Lena, 27 — customer service agent
| Aspect | Detail |
|---|---|
| Context | Drafts customer replies daily with the new AI assistant at Northwind Logistics |
| Prior AI use | Daily, uncritical — trusts fluent output |
| Key risk | Pastes customer data into prompts; sends unverified claims (“we always deliver within 48 hours”) |
| What she needs | A fast verification routine that fits her handle time; clear data rules |
“The AI writes better than I do, and it’s never wrong in an obvious way.”
Design implications: Module 2 must give a checklist that takes seconds, not a philosophy of epistemology. Module 3 must give a bright-line rule she can apply under time pressure. Quiz items for her use customer-reply scenarios with time pressure in the stem.
Joris, 45 — operations planner
| Aspect | Detail |
|---|---|
| Context | Uses AI to summarize weekly operations and delivery reports |
| Prior AI use | Occasional, skeptical |
| Key risk | Trusts numbers that appear in AI summaries — they look authoritative but may be invented |
| What he needs | Knowing where AI fails with numbers; a concrete check for figures |
“I don’t trust the chatbot’s opinions. Numbers are numbers though.”
Design implication: Module 2 needs an explicit “AI and numbers” section: invented figures, wrong totals, plausible-looking tables. His practice exercise is a summary with three fabricated numbers.
Sofia, 38 — HR administrator
| Aspect | Detail |
|---|---|
| Context | Considering AI to screen incoming CVs |
| Prior AI use | Low |
| Key risk | Bias against groups; decisions about people delegated to a system; unclear legal context |
| What she needs | Recognizing high-risk use; knowing who decides; an escalation path |
“If it saves the team a day of screening, why not?”
Design implication: Module 4 centers on her scenario (the branching mini-scenario “The HR shortcut”). The content must avoid legal jargon but name the principle: the more a decision affects a person, the more a human belongs in the loop.
School track
Ayo, 15 — student
| Aspect | Detail |
|---|---|
| Context | Uses chatbots for homework several times a week |
| Prior AI use | High, fluent |
| Key risk | Offloading thinking; trusting fluent text; not understanding why answers are wrong |
| What he needs | To see how the text is produced (prediction), and rules for when AI helps learning |
“It just knows things. If it’s wrong, why does it sound so sure?”
Design implication: Lesson 1 must produce a genuine “aha” (he makes the model choose next words in the interactive). Lesson 4 must not moralize — it co-designs rules with students.
Mr. Daniels — teacher (non-specialist)
| Aspect | Detail |
|---|---|
| Context | Teaches the unit as part of the curriculum; IT background is basic |
| Prior AI use | Low–medium |
| Key risk | Low confidence; fears student questions he cannot answer |
| What he needs | A ready-to-teach guide with answers, timings, and “if a student asks X, say Y” support |
“I can’t answer technical questions about AI. I need the answers in front of me.”
Design implication: Every lesson plan includes a misconception table with suggested responses, exact timings, and an unplugged fallback. Nothing requires the teacher to improvise technical content.