Work With Us - Every deployment starts with a conversation

ConductVision is designed to work with any camera that supports RTSP or ONVIF protocols, which includes the vast majority of IP cameras from manufacturers like Axis, Hikvision, Dahua, Bosch, Hanwha, and many others. USB and CSI cameras connected to edge devices are also supported. If your camera can produce a video stream, our technology can likely process it.

A typical pilot runs 4-8 weeks in your environment with your cameras and your data. We start by understanding your use case, then deploy our detection models on-premise, configure dashboards for your specific needs, and measure performance against agreed success criteria. At the end, you get a detailed report with transparent accuracy metrics so you can make an informed decision about full deployment.

ConductVision is built on a decade of ConductScience experience in laboratory animal behavioral tracking, where our technology is proven and deployed in research institutions. We are actively expanding into new industries and seeking partners who want to bring science-grade AI vision to their domain. Our detection models are built and tested — we are looking for the right environments to deploy them.

All processing is designed to run on-premise on your hardware. Video streams are processed in real time, and only event metadata (timestamps, bounding boxes, alert types) and short event clips are stored locally. Full video streams are not recorded unless you explicitly enable clip recording. All data stays within your network — nothing is sent to external servers.

ConductVision is designed for fully on-premise deployment. We provide a containerized system that runs on your own hardware, within your network — no video or data ever leaves your infrastructure. The deployment includes detection models, the dashboard, and the API. For multi-site organizations, each location can run its own instance with centralized management.

Yes. Custom model training is available as part of pilot and deployment engagements. You provide labeled example images of the objects or conditions you want to detect, and we train a model optimized for your specific use case using transfer learning. This is one of our core strengths — we have deep experience training and validating detection models from our research heritage.

We work best with organizations that have existing camera infrastructure and a specific operational challenge they want to solve with AI vision — whether that is safety monitoring, quality inspection, animal welfare, or operational efficiency. We value partners who appreciate scientific rigor and transparent metrics over marketing hype. If you have cameras and a problem, we want to hear about it.

Bandwidth requirements depend on your camera resolution and frame rate. As a guideline, a 1080p stream at 15 FPS requires approximately 2-4 Mbps. ConductVision can also process lower-resolution sub-streams (720p or lower) for many detection types, reducing bandwidth to under 1 Mbps per stream. We assess your network capacity during pilot scoping.

A discovery call takes 30 minutes and we can usually schedule one within a few days. From there, pilot scoping typically takes 1-2 weeks, and pilot deployment takes another 1-2 weeks depending on your environment. Most organizations go from first call to running pilot in under a month.