What this case covers
Using Bilibili e-commerce as the scenario, the author reviews AI collaboration methods from product selection and requirement collection to video production, and explains how to build a project knowledge base and a repeatable operating system.
Questions it answers
- How should humans and AI divide responsibilities?
- Why build a project map first?
- How are comment-section requirements turned into product selection criteria?
- How to accept 70-point results and keep correcting them?
- How does Token investment settle into assets?
Our observations
- Model usage is not capability; closed-loop iteration is.
- Humans handle goals, responsibility, and aesthetics; AI handles repetitive execution and self-checking.
- Product selection content should reduce users' choice cost, not pile up specifications.
From the author
Humans handle goals, boundaries, responsibility, and aesthetics; AI handles data processing, repetitive production, self-checking, repair, and delivery.