What we do
An engagement, not a product on a shelf.
The components are inexpensive and the software is open. What costs time is everything between hardware on a desk and a loop trusted to run overnight. We do that part and hand over something with numbers attached.
1. Work out what it has to drive
Which machines, which screens, and whether the job needs AMI at all
rather than an API that already exists. If an API will do, we say so.
2. Pick the approach and prepare it
Chosen against how fast it has to be and what it has to reach. We supply it
ready to use.
3. Get it working on the target
Every machine accepts a keyboard and gives up a picture slightly
differently. This is where most of the difficulty lies, and it is the work
the engagement is for.
4. Connect it to the agent
Joined up to whatever the operator is already running, so the agent can
use it without anything being rebuilt around it.
5. Measure it on site
Frame rate and action-to-visible taken on the operator's hardware, by the
same method as our own numbers: what is actually there, not what we
claim.
6. Hand it over
A runbook for the failure modes we hit on that equipment, and the
measurements to re-run when something changes.