The vision model behind InvManage.
A cloud-hosted computer vision pipeline applied to the photos your team already takes — no on-site hardware to install, power, or maintain.
Computer vision, run in the cloud
InvManage runs object-detection and vision-language models on the photos your staff upload — no cameras to mount on your racks, no edge boxes to provision. A photo goes up, the model scores it, and a finding comes back with a confidence level and a plain-language explanation.
Because we're an early-stage product, the model is still being tuned against real warehouse photos from our pilot operators. We're upfront about that: some flags will need correcting, and every correction feeds back into making the model better.
The cloud stack behind it
Everything below runs in the cloud, on infrastructure we rent from AWS, Azure, and NVIDIA — not equipment installed at your facility.
NVIDIA cloud GPUs
Used for training and running inference on our detection models — rented cloud compute, not hardware shipped to your site.
Azure Cosmos DB
Stores flagged findings, review history, and confidence scores with low-latency reads for the review queue.
Azure AI Vision
Provides pretrained vision building blocks our detection models build on top of, alongside our own fine-tuned models.
AWS S3
Stores and backs up the photos your team uploads, with access scoped to your organization's account.
Why not edge hardware?
Rack-mounted cameras and edge processors are expensive to buy, slow to install, and break the moment a warehouse's layout changes. We deliberately built InvManage the other way: point the model at photos people are already taking, and let the software do the work.
Nothing to install
You can start uploading photos the same day you sign up — no site visit, no installation crew, no downtime.
Works anywhere your team can walk
A new aisle, a rented overflow facility, or a supplier's dock — if someone can take a photo there, the model can screen it.