Development continues after competition feedback
Yangtse Evening Post's October 12 report describes student entrepreneur Ding Yi's Plot Cube and competition experience. It explicitly says she plans further agent and interactive development, aiming for a mature usable product next year. A competition work and an ongoing user product are different stages.
This early content-tool example asks who controls direction, what AI generates, how people edit and whether interaction remains coherent. Our prototype suggestions are not current product specifications.
Select a user before choosing story length
Creators may want to compare directions, while audiences want meaningful choices. Start with one situation, such as limited branches for an original short story, rather than every genre and interaction.
In a simulated writer submission, confirm characters, world rules, duration and permissible changes before generating a few branches. Tell testers whether they are evaluating creative assistance or consuming finished content so feedback does not mix incompatible expectations.
Check branches as connected paths
A plausible paragraph may contradict identity, forget clues or reveal an ending when connected. Test complete paths from entry through different choices, checking events, retained information and broken rules, not only isolated prose quality.
Maintain a story-state table of character knowledge, events and available actions. AI suggests alternatives; creators approve versions and record reasons, helping distinguish rule, input and generation problems.
Tie feedback to specific choices and passages
A dull-story comment is insufficient. Ask which choices lacked difference, where information overloaded or behavior became implausible. Decide whether to change narrative, interface or pacing rather than translating every problem into a model replacement.
Explain trial-material uses and retain sources, licenses and contributions. Generation does not remove underlying character or material rights. Establishing records during prototyping is easier than reconstructing them after launch.
Let stage-specific evidence guide investment
Initially observe operation, path completion, retry interest, errors and revisions. Later study creator retention, support costs and accepted prices. Competition judgments cannot replace these distinct forms of evidence.
The report does not disclose a legal entity, current headcount, paying users, orders or a mature product we can verify. Awards are not commercial success. Track the progression from an idea to a checkable experience without portraying future plans as completed results.
Further reading: AI video delivery; case corrections. Haizhou Wukong Zhihua Community.