Yidan Wang — Business Management × Computer Science × AI
Driving Business Decisions with Data.
王怡丹 —— 商业管理 × 计算机 × AI，用数据驱动商业决策。
“Case teaching gives you the first opinion. I use data to deliver the second.”
“案例教学给的是第一意见，我用数据给出第二意见。”
Read the case → write the analysis → discuss in class → professor comments → class ends. Across hundreds of cases, a student never once hears “you were right” or “you were wrong.” Practice without feedback does not improve — this product fixes exactly that one link.
读案例 → 写分析 → 课堂讨论 → 教授点评 → 下课。几百个案例讨论下来，从没有人告诉她“你对了／你错了”。没有反馈的练习不会进步——这个产品只补这一环。
A simulation of reality. Many possible answers, no stated conclusion — inference as the primary skill.
对真实情况的模拟：答案不唯一、案例不给结论，推理是第一技能。
What happened next in the real world. Whether the decision actually worked. A verdict.
真实世界里这个决策后来怎么样了；当年做没做对；一个裁决。
She decides. Data referees.她做决策，数据当裁判。
Easy first: bilingual cases, familiar brands, single decision points lead; industry distance, data complexity and ambiguity rise tier by tier. One case per week in term time, two per week over the long summer and reading breaks — all 120 cleared before Ivey, September 2028.
简单的排前面：中英对照、熟悉品牌、决策点单一者先行；行业陌生度、数据复杂度、决策模糊度逐层上升。学期中每周一案，暑假（5–8 月）与阅读周每周两案，2028 年 9 月进 Ivey 前全部跑完。
Demonstrated with Case 000 — Tim Hortons: Bringing an Iconic Canadian Coffee to China. Step through the loop as a player would, or play it live on the ladder wall.
以第 000 案《蒂姆·霍顿斯：把加拿大国民咖啡带到中国》为例演示。点击四个步骤速览，或到阶梯墙里真打一局。
DEMO WITH PUBLIC RECORDS ONLY 演示仅使用公开记录（SEC 文件、公开报道），案例原文不分发。
The scoreboard is not a feature — it is a by-product. Every case that completes the loop writes one entry; 120 cases over two years compound into her decision-accuracy curve.
记分牌不是功能，是副产品：循环每跑一案记一笔，120 案 × 两年自然攒出她的判断准确率曲线。
SAMPLE TRAJECTORY 示意轨迹（真实曲线自第 001 案起跑后生成）
记录含每案的选择、把握度、训练分、用时与完成时间。
Early cases need only one strong model behind the referee. As the ladder climbs, cross-validation kicks in — ChatGPT, Claude and the leading Chinese models debate in rounds, disagreements go deep, and “Option D: defer the decision” always stays a legal answer. All backstage, invisible to the user.
前期案例只需一个强模型当裁判；阶梯往上走，再上交叉验证——ChatGPT、Claude 与中国领先大模型分立场辩论，分歧点深挖，“选项 D：暂不决策”永远是合法答案。全部藏在后台，界面不可见。