AI NPCs
Suspects answer from their identity, known clues, and pressure state instead of fixed dialogue options.
misty.ngjoo.com · EXPERIMENT
Misty Harbor
A disappearance at a Republican-era theatre. Interrogate AI-driven suspects in natural language and collect evidence from evasions, changed stories, and contradictions.
Game mechanics
COREGated clues
Verify timelines before pressing motives and opportunities; key evidence appears only in the right confrontation.
Evidence-based verdicts
Evidence quality determines whether you continue, accuse incorrectly, or reach the full ending.
Who it is for
Useful for players who want a narrative experiment built around natural-language interrogation, gated clues, and evidence-based verdicts.
Read the boundary first
This is a fiction experiment, not an investigation tool. AI character responses belong to game state and are not factual testimony.
Ways to use it
- Reconcile each suspect's timeline and group statements that corroborate one another.
- Use acquired clues to press changed stories or contradictions instead of traversing dialogue at random.
- Review evidence quantity and relevance before accusing, returning to incomplete branches when needed.
How to review the result
- Character responses depend on identity, clue state, and pressure and only hold inside the fiction.
- A missing clue may mean that a prerequisite was not met; it is not necessarily a negative finding.
- Different dialogue paths can reveal information in a different order, so review it with the current save state.
Current status and evidence
2026-09-23The endpoint redirected to NGJOO sign-in on 2026-09-23; deployment records keep it inside the pilot scope.
The available evidence is first-party: deployment records, product documentation, and endpoint status at the review time. No independent evaluation was available. Reachability confirms access at that moment, not performance, accuracy, savings, or long-term availability.