Research program · Ο · Sight

Making a place knowable.

Sight is the awareness spine — the program that turns the constant weather of a place's signals into one honest picture, and then turns that picture into a plan. It is the single clearest answer to the question every visitor asks: what do you actually do? We make places knowable, and we make knowing useful.

All programs Talk to us →
The idea

Awareness that plans, not just watches.

A place — a hospital, a base, a border stretch, a warehouse, a district's clinics — produces a constant weather of signals: cameras, sensors, logs, readings, movements, absences. Almost everywhere, that weather goes unread, or is read too late, or is read only by exhausted people. Sight stands a quiet system inside that weather and keeps it answerable at any moment.

The program's discipline is that awareness which only watches is a dashboard, and dashboards are ignored within a week. Awareness that plans is a colleague. So Sight is built to answer not only what is happening now, but what is about to happen, what matters most out of everything happening, and who should act — and then to look further ahead, to the service a machine will need this month and the infrastructure a district will need in five years.

Sight deliberately does not chase the biggest general model. It selects, shapes, and grounds the best available intelligence for a specific place and task, and it must run on both compute tracks — the advanced one and the plain, locally-buyable one. Small, sharp, local, and durable beats large, general, distant, and fragile. A warning nobody can read, on a system that goes dark when the network drops, was never really a warning.

Inside the program

What Sight is working on.

01

Sensor fusion on cheap hardware

Taking whatever a place already owns — old cameras, mismatched sensors, paper-era logs — and fusing them into one coherent picture on a small computer at the edge. The research is in doing this well on modest silicon, over weak or absent connections, without a cloud to lean on.

02

Plain-language priority

Turning a flood of detections into a short, ranked, human-readable brief. The hard part is not detection; it is deciding what a busy person needs to hear first, and saying it in words that need no training to understand.

03

From watching to planning

Extending awareness forward in time: predicting which equipment will fail, which stock will run out, which corridor will be crowded, and what a place will need seasons and years ahead — so growth is planned rather than improvised.

04

Honest uncertainty

Teaching the system to say how sure it is, and to flag when it is guessing. A system that hides its own doubt is more dangerous than one that admits it, especially where a human is about to act on what it says.

Now · ~1yr

What runs today

Awareness spines live in first hospitals, clinic networks, and protected sites, fusing existing sensors into one plain picture with sorted priorities.

Next · 1–2yr

What comes next

multi-site pictures for organizations with many locations; deeper planning features; voice interaction matured across more languages.

Later · a bet

A genuine research bet

awareness that plans a whole district's infrastructure years ahead, and reads intent from patterns across many quiet signals. We label this honestly as a bet, not a promise.

Keep exploring

Where Sight shows up

See what your place has been trying to tell you.

Sight is where our awareness work lives. Tell us about the place you need to know.