Interactive Widget Demo
Core Sentinel as an independent floating desktop overlay. The browser window below is only backdrop context; the widget hovers above everything.
Interactive Widget Demo — click a scenario or interact with the overlay directly
Mode A focuses on warning/block behavior and remediation actions.
How it works (current phase demo)
Idle: choose a scenario
1) Paste intercepted
2) Risk scan runs
3) Safe/Warn/Block decision
4) Remediation actions
5) Safe version proceeds
Tour idle
Source Document
LLM Prompt
https://llm-workspace.local/prompt-lab
Copied from document
Use the quick-copy buttons to simulate selecting text in a doc and pasting it into the LLM prompt.
Quarterly Incident Notes
Owner: Priya Shah
Email: priya.shah@acme.org
Phone: +1 415-555-1177
Summary: Need safer prompt before sharing externally.
LLM prompt box and paste log
Paste content from your document to see how Sentinel intercepts and classifies it.
⏻
◈
CS
Monitoring: ChatGPT
SILENT
Start a streak!
Monitoring: ChatGPT
Start a streak!
🪟 Start
🌐 Chrome
🟦 VS Code
📁 Explorer
12:30 PM
🔔
Select document to scan for PII
employee_records.pdf14 KB
config_backup.txt2 KB
patient_notes.docx8 KB
Document Scan Report
Retrain model?
You have 23 corrections collected.
Retrain the model now? (takes ~2 minutes)
Overlay tour
Start tour to see each part of the widget.
This is a browser simulation. The real Core Sentinel desktop app provides:
- 🧠 TinyBERT guardrail model for contextual risk scoring (not just regex)
- 🧬 spaCy NER for name/org/location entity recognition
- ⚡ System-level clipboard interception (works in ALL apps)
- 🔒 AES-256 encryption for reversible PII protection
- 📊 Supabase-powered fleet telemetry for enterprise
- 🔄 Active learning: model improves from your corrections
- 🪟 Current production target: Windows 10/11 (Linux and iOS planned later)
Download the real thing →