Practice simulators
Skills break down in conversations, so these simulators let people rehearse them. Each one puts the learner in a realistic situation with an AI character who has hidden motives, then scores the learner against a five-part rubric using evidence from what they actually said.
Work sample by Allan DeYong, built on the OpenAI API. All companies, people, and documents are fictional.
The AI skeptic
Win over an operations manager who has postponed her team’s AI rollout twice. Find the real problem, handle the accuracy risk, and agree on a pilot.
Start → AI skills · Prompt writingFrom SOP to job aid
Write a prompt that turns a messy warehouse procedure into a one-page job aid. Run it, see the output, and get scored on both the prompt and the result.
Start → Leadership · Coaching conversationsThe slipping deadline
Give feedback to a strong employee whose work has slipped. The real reason only comes out if you ask and listen.
Start → Frontline · Guest serviceThe wrong order
A regular guest waited 25 minutes for an order that’s missing half the food. Recover the moment, and the relationship, as the shift leader.
Start → Sales · Account managementThe renewal conversation
Move a skeptical law-firm IT director toward the cloud at renewal by uncovering what’s really driving her decision.
Start →How they’re built
Hidden motives
Each character has stated objections plus needs they reveal only when the learner earns them, so a memorized script doesn’t work. Listening does.
Evidence-based rubrics
Five observable behaviors per scenario, scored 1–4 from the transcript. Feedback quotes what the learner said and gives one specific thing to try next.
Guardrails
Prompts stay on the server. Per-visitor and daily limits keep costs predictable, and the site doesn’t store conversations.