SafeX Insight is a data-analytics-driven public-welfare app (mobile H5) that helps women facing harassment, stalking, domestic violence and other safety threats understand their situation, collect evidence, and get help fast.
Problem it solves: victims often can't tell whether what happened is illegal, don't know what evidence to keep, where to seek help, or how to apply for a personal safety protection order. Fragmented data makes the real scale and geographic patterns of women's safety incidents invisible.
Target users: women experiencing harassment, stalking, workplace/sexual harassment or domestic violence in China, as well as their families and public-interest/legal-aid workers.
Core features:
1. Safety data insight dashboard — type distribution, 24-month trends, province TOP10, help channels and outcome ratios across 2,300+ incident records, with an Amap point/heatmap view and linked multi-condition filters.
2. AI assistant + AI incident analysis reports — users describe what happened conversationally and receive a structured report: legal qualification, data interpretation, similar cases, legal basis (real statute citations), action advice and an evidence checklist.
3. SOS six-screen guardian flow — nearby aid resources map, long-press 3s alert, live GPS guardianship, one-tap 110 dialing and platform reporting.
4. Safety toolbox — fake call, timed guard, illegality self-assessment, protection-order wizard, 5-step safety plan, safety diary, abuser profile, 7-step help progress tracker, PIN lock and one-tap emergency exit.
5. Content library — 32 typical cases, 42 legal Q&As, 15 law-education articles and 130 geo-tagged aid resources.
How it uses the InfiniSynapse API:
- AI analysis: the backend (Java 17 BFF) creates async tasks via POST /api/ai/message (newTask), matches the task via GET /api/ai_task/list, and polls GET /api/ai_task/tasks to extract the structured report, using the safety-incident dataset uploaded to the platform as its analytical context.
- Data sync: on startup it idempotently creates two file-type data sources (safex_events, safex_statutes) via POST /api/ai_database/add and uploads exported CSVs via POST /api/ai_database/upload/:id, so judges can verify both data sources and AI tasks in the platform console.
- Multi-level degradation: if no API key is configured or any call fails, the app automatically falls back to a local template engine backed by real SQLite queries, so every feature still works end-to-end.