Experiments
AI-generated content business tests
This page tracks AI-native content experiments as business lines. Each one has a format, an audience hypothesis, a measurable signal, and a next production decision.
Portfolio
Active and staged lines
Page views are not enough. A line needs repeat visits, completion, comments, follows, downloads, CTA clicks, or purchase intent before it earns more production time.
Aila Trace
An AI-generated investigator-idol character designed for story shorts, case files, avatar packs, and future digital goods.
- Validation metric
- Profile views, case clicks, avatar interest, short-video follow-through
Approval Queue
A browser-native decision game that turns AI agent approval risk into a playable review loop.
- Validation metric
- Completion rate, replay intent, decision feedback, service CTA clicks
The Refund That Approved Itself
The first Aila Trace incident case, built to test whether AI workflow failures can become repeatable story content.
- Validation metric
- Case reads, profile clicks, story completion, follow-up CTA clicks
The Human Approval Layer
A near-future serialized story world about human approval, broken automation, and systems that optimize the wrong metric.
- Validation metric
- Read-through, saves, comments, subscriptions, return visits
Operating rules
Generated content must prove audience signal
AI can produce more content than the audience can care about. This lab favors recognizable characters, consistent worlds, reusable formats, and small releases that can be measured quickly.
- Log every generated asset source and reuse rule.
- Ship one small public artifact before scaling a format.
- Separate metrics by line: IP, story, game, video, and digital goods.
- Stop formats that get views but no completion, return, or action.