Percival captures a frame every 8 seconds, describes it using a tiny on-device vision-language model (VLM), and sends text — not video — to the PerceptEdge cloud. A powerful large language model (LLM) analyzes patterns and detects potential risks.
There is no continuous video streaming, no 24/7 remote surveillance. Only meaningful information is transmitted when it matters.
Percival provides care, not monitoring.
Percival is designed for people and organizations that need timely awareness without intrusive cameras:
• Older adults living alone
• People with limited mobility
• Rehabilitation patients
• Social care centers and assisted living
• Municipal or NGO assistance programs
Percival enhances safety without intruding on everyday life or turning a home into a surveillance zone.
✔ Respects privacy – only sends text unless an event occurs.
✔ Extremely low bandwidth — works even on weak Wi-Fi or metered connections.
✔ No technical maintenance required for everyday use.
✔ Designed for large-scale social and municipal projects.
✔ Customizable behavior through an AI prompt-agent instead of complex re-training.
✔ Affordable and easy to deploy in real homes, not just pilot labs.
Percival is not a camera. It is an intelligent helper that notices when something may be wrong — and makes it easy to respond in time.