Workplace security monitoring, built with privacy in mind.
Founding Minds engineered Proktor, an AI-powered workplace monitoring platform designed to help organizations protect sensitive information in remote and hybrid work environments. Using edge-based computer vision, targeted event detection, and privacy-conscious image handling, the system provides visibility into defined security risks while limiting unnecessary employee observation.
The situation
As work moved out of the office, sensitive information moved into home environments, outside the physical controls of a supervised floor. For financial services, healthcare operations, legal services, and other regulated organizations, that created new information-security concerns.
Proktor is a JIFFY.ai product. Founding Minds engineered it.
What made it hard
Identifying specific events without indiscriminate observation.
The system was designed to look for defined security events instead of capturing everything happening around a remote worker. Deciding what is detected, and nothing beyond it, shaped the rest of the design.
Keeping inference close to the endpoint.
Processing video locally reduces latency and the need to transmit raw video to a remote service. What is stored, logged, and passed to the monitoring interface is a separate question that the architecture had to address deliberately.
Handling variation in home workspaces.
Webcams, lighting, and room layouts vary from home to home, which makes real-world computer vision harder than a controlled demonstration. The models were designed to adapt to that range of conditions.
Integrating with enterprise environments.
The target customers are established organizations with existing infrastructure and security workflows, including legacy systems. The architecture is modular so it can integrate without requiring those environments to be modernized first.
What was built
An endpoint-monitoring system, in three parts.
Edge-based computer vision
Models run on the endpoint and detect configured security-related events, adapting to variation in camera quality, lighting, and room layout.
Privacy-conscious image handling
Image anonymization techniques reduce unnecessary exposure of identifiable employee imagery.
Event monitoring and enterprise integration
Presents relevant detections to authorized teams, through a modular architecture that integrates with existing enterprise IT, including legacy infrastructure.
- A phone in the camera view
- An additional or unrecognized person in the workspace
- A sensitive document visible on screen
- Screenshot activity
Defining the events makes the system easier to assess: a security team can state what it looks for.
What changed
Targeted security visibility for distributed work environments
Proktor brings configured security-event detection and monitoring into a system designed for remote and hybrid workplaces. Its use of local processing and privacy-conscious data handling reflects the need to balance security visibility with limits on employee observation.
The engineering decision that mattered
The team treated privacy constraints as part of the system architecture rather than something to address after development. Local inference, targeted detection, and careful handling of visual information shaped how the product was designed and integrated into enterprise environments.