You know what is inside
Every part lot, firmware build and AI model on a device is traceable to its serial number. No unknown firmware, no undocumented cloud calls, and a clear answer when procurement or audit asks where the device came from.
IndoAI designs, assembles, flashes and tests its edge AI hardware in house. Our cameras and the EdgeBox run on the NVIDIA Jetson platform, carry our own firmware and AI runtime, and are signed off unit by unit before they leave the floor.
Two hardware products carry the IndoAI name. Both are engineered by our own hardware and software teams, which is why the same AI agents run on either of them without rework.
A camera with the computer inside. Sensor, lens and a Jetson module share one sealed housing, so detection runs in the camera itself and only events and alerts leave it. Built for new sites and for single camera locations where a separate box makes no sense.
The retrofit unit for sites that already have CCTV. It sits on the local network, pulls RTSP streams from the cameras or NVR you own today, and runs the AI models on site. Nothing gets ripped out, and no video has to go to a cloud to be analysed.
Where one site has many cameras, we configure, image and burn-in test a single Industrial PC for that site. It goes through the same release process as the cameras and the EdgeBox, runs the same agents from the Appization marketplace, and shows up in the same app.indo.ai console.
Six stations, one serial number. Every camera and EdgeBox moves through the same sequence, and the record of each station travels with the unit.
Enclosure, thermal path, board and I/O are engineered by our hardware team and frozen as a controlled build document.
Jetson modules, sensors, lenses, storage and power parts come from qualified vendors. Each lot is inspected before it reaches a bench.
Boards, modules, optics and housings come together at our Pune benches. Cameras are focused and sealed at this station.
The unit receives the IndoAI system image, its security keys and the AI agents ordered for that site.
Units run live video under full inference load. Thermal, network, storage and detection checks are logged against the serial number.
The nameplate goes on, the device is registered on app.indo.ai so it appears in your fleet on first boot, and it ships with its test record.
Most AI video products are software loaded on someone else's imported hardware. When the hardware, firmware and models come from one company, a few things change for the buyer.
Every part lot, firmware build and AI model on a device is traceable to its serial number. No unknown firmware, no undocumented cloud calls, and a clear answer when procurement or audit asks where the device came from.
Hardware, firmware and AI sit under one roof. A field issue is not passed between a camera vendor, a box vendor and a software vendor. The people who built the unit are the people who fix it.
Lens, housing, I/O, preloaded agents and branding can be set at build time for system integrators, OEM and white label partners.
Spares, repair and replacement run out of Pune, with firmware and model updates for the life of the deployment.
Inference runs on the device. Video does not leave your premises unless you decide it should.
Nobody builds a camera from sand. Here is the honest split between the components we buy and the engineering, assembly and testing we do ourselves.
| Part of the product | Where it comes from | What IndoAI does |
|---|---|---|
| Compute module | SourcedNVIDIA Jetson | Module selection per product, thermal design around it, validation under sustained load |
| Image sensor and lens | SourcedQualified vendors | Selection per use case, focusing, image tuning for the AI models that will read the picture |
| Housing and heat sink | IndoAIOur design | Mechanical and thermal design, assembly and sealing |
| System image and firmware | IndoAIBuilt in house | Build, hardening, device keys and over the air updates |
| AI runtime and agents | IndoAIWith Appization developers | Optimisation for Jetson, packaging, versioning and delivery through the Appization marketplace |
| Final assembly and test | IndoAIPune | Assembly, flashing, burn-in, serialisation and dispatch |
The nameplate on an IndoAI device is more than a sticker. The serial number on it opens the full build history of that unit.
Sample nameplate for illustration. Serial format shown is an example.
Each device ships with the log of its own burn-in and functional test, not a batch sample certificate.
The serial number links to the part lots used, the firmware build flashed, the agents loaded and the person who signed it off.
BIS registration for the EdgeBox range is under way. We will publish the registration details here once granted.
Device keys are provisioned on our floor, updates are signed, and the unit works fully offline for sites that need it.
Buyers, system integrators and white label partners are welcome on the floor. Walk the line, open a unit, and talk to the production and engineering teams about volumes, custom builds and lead times.