ContactMenu
AIT Lab

Edge AI for places the cloud does not reach.

Small devices that decide on site, keep their own records and send them on when the network returns. This is lab work. Each item shows the stage it has actually reached.

Tweezers place a small chip on a green circuit board.
  1. A solar panel canopy on a building in a dry rural town.
    H1

    Energy security

    PROTOTYPE

    Edge sensing for distributed energy systems. A small board reads current and voltage on each source, decides on site when something is wrong, and keeps every reading until it can send them.

    The problem
    Sites switch between grid, generator and solar through the day, and nobody records what each source delivered, when it failed or what it cost.
    Design constraint
    Has to run through outages on its own battery, store readings when the network drops, and be repairable with parts sold in Lagos.
    What we are looking for
    • A test site
    • A co-development partner
  2. Frozen fish stacked in metal cages in cold storage.
    H2

    Food security

    CONCEPT TO PROTOTYPE

    Environmental sensing for cold rooms, with spoilage prediction. An independent sensor watches temperature and humidity, so a fault or a tampered controller cannot hide a rising temperature.

    The problem
    A cold room that drifts a few degrees overnight can spoil a week of stock before anyone opens the door.
    Design constraint
    Must work independently of the cold room's own controller, raise an alarm by SMS when the network is weak, and cost less than the stock it protects.
    What we are looking for
    • A test site
    • Grant co-applicants
  3. Network cables plugged into a router with status lights.
    H3

    Cybersecurity

    CONCEPT

    Edge monitoring for local networks. A small device on the network lists what is connected, notices what changes and flags the devices that are open to the internet.

    The problem
    Small businesses run routers, cameras and connected equipment that nobody watches, on networks that nobody has mapped.
    Design constraint
    Has to run on a cheap board inside the network, need no cloud subscription, and report in plain language.
    What we are looking for
    • A co-development partner
    • Grant co-applicants

Have a site where one of these could be tested?

A cold room, a generator house, a small office network. Real conditions are what the lab needs most.

Partner on a pilot