AutoSense™ is the product of Pujaso's view that the future of security is not more cameras — it is more intelligence applied to the cameras you already have. The platform was designed around a single question: what would it look like if every data source in a security deployment — cameras, access control, IoT sensors, fire alarms, environmental monitors — was connected to a single AI brain that understood context, identified patterns and acted automatically?
The Five-Layer Processing Pipeline
Layer 1 — Ingest: AutoSense™ connects to data sources via RTSP streams (cameras), REST and WebSocket APIs (access control, building management), MQTT brokers (IoT sensors), SNMP (IT infrastructure) and direct database connectors (ERP systems, HR platforms). Every event from every source enters the unified data pipeline in real time.
Layer 2 — Normalise: Data from dozens of different vendors and protocols is normalised into a common event schema. A door open event from a Honeywell access panel and a door open event from a HikVision panel are represented identically in the AutoSense™ data model — allowing cross-system correlation without custom integration code for each combination.
Layer 3 — Analyse: The core AI processing layer. Video streams are processed by object detection, facial recognition, behavioural analytics and anomaly detection models running in parallel. Non-video events are processed by classification models that assign event type, severity and entity associations. This is where raw data becomes structured intelligence.
Layer 4 — Correlate: AutoSense™'s correlation engine applies temporal and spatial reasoning across normalised events from all sources. It identifies patterns that span multiple systems — a camera anomaly followed by an unusual access event followed by an IT login outside business hours — and scores composite threat indicators with explainable confidence metrics.
Layer 5 — Act: High-confidence threat assessments trigger configured response playbooks. Playbooks are visual workflow sequences that define what should happen: who gets notified, what cameras should PTZ to, which doors should lock, what systems should be disabled, what reports should be generated. Playbooks can be fully automated for high-confidence event types or require human approval for ambiguous situations.
Integration Architecture
AutoSense™ connects to 50+ systems out of the box. The integration philosophy is non-disruptive: the platform reads from and writes to existing systems without requiring replacement. Your existing Hikvision NVRs, your HID access control system and your fire alarm panel continue operating independently — AutoSense™ acts as an intelligence overlay, not a replacement.
Deployment Models
Cloud-hosted SaaS: AutoSense™ runs in Pujaso's hosted cloud infrastructure. Ideal for organisations without on-premise server capacity or IT teams. All processing, storage and backups are managed by Pujaso.
On-premise enterprise: AutoSense™ deployed on customer-owned hardware inside the facility. Required for environments with data sovereignty requirements, low-latency needs or air-gapped network policies.
Hybrid: Edge processing units at each site perform time-sensitive local analysis; a central cloud or on-premise server handles cross-site correlation and long-term analytics. The most common enterprise deployment model for multi-site organisations.