Sentinel prototype
公開証拠を権力のループとして可視化。
Sentinel turns public records, official sources, historical documents, and market/policy data into evidence-graded loops, showing what is connected, what is missing, and what supports each link.
What is a loop?
Sentinel shows how people, companies, money, policy, and events may connect - and where the evidence is strong, weak, missing, or contested.
Try now
Start with live prototype surfaces.
These are artifact-backed demos. They show the product model without live AI, live ingestion, or hidden internal routes.
Oct. 7 Event Map
Multiple loops intersect around the Oct. 7 event node. Sentinel separates the modern defense money-policy-benefit loop from the contextual Iran/proxy loop and grades the evidence differently for each.
Defense Money-Policy-Benefit Loop
A partial loop connecting defense companies, PAC money, vote-connected officials, the Israel/security supplemental policy path, and market exposure. The company-specific benefit and insider/stake links remain missing.
Iran / Proxy Contextual Loop
An open contextual loop around Iran policy, sanctions-relief arguments, proxy networks, Hamas/Hezbollah distinction, and the broader Oct. 7 conflict. Contested links remain clearly labeled.
Iran-Contra Historical Scaffold
A historical retrospective scaffold showing how Sentinel will evaluate older events using investigation reports, court records, declassified materials, testimony, archives, journalism, and historical analysis instead of modern-only datasets.
Solar in Texas Event Seed
A lightweight event seed showing how Sentinel can remember weak signals cheaply before promoting them into full investigations.
How Sentinel scales
Watch broadly, go deep selectively.
Sentinel does not research the whole world all the time. It watches broadly, remembers weak signals cheaply, and launches deeper loop analysis only when a signal strengthens, a user asks, or a historical backfill is triggered.
Daily watch
Finds weak signals and creates event seeds.
Triggered investigation
Expands a user question into scoped loop analysis.
Backfill
Rebuilds historical events as better data and better models become available.
The product is built around restraint.
Sentinel is designed to show evidence posture before conclusions. Open links are product output, not failure.
Every link carries evidence status
Missing links are shown, not hidden
Modern loops require modern structured evidence where available
Historical loops use retrospective official and archival records
AI proposes candidate edges but does not decide truth
No loop closes without source-backed evidence