Rhode Island · United States

Applied R&D.
Operational advantage.

OSAT turns research-grade machine intelligence into systems that can be fielded — for U.S. industry and government partners who cannot afford theater.

PracticePerception, sensing, decision systems
ProductCIPHER — fragmented & distributed detection
LeadershipEamon Conway, CEO

Why we exist

Most AI never leaves the lab. That is the gap we close.

We are a consulting firm with a product spine. Clients hire us to take methods from computer vision, geospatial sensing, and network science and make them work in the environments that actually matter — austere, real-time, and unforgiving of false confidence.

Practices

Three sectors. One standard.

We do not sell a buffet of buzzwords. We run three practices, each with a named problem and a way of working.

01 — Perception & autonomy

See, locate, act

Object detection, 3-D localization, scene understanding, GPS-denied navigation, and autonomous systems that have to decide without a perfect map.

Open the practice →

02 — Sensing & the physical world

Read the terrain

Remote sensing and large-scale geospatial analytics — turning spectral, radar, and location data into something an operator can use.

Open the practice →

03 — Decision systems

Structure the hidden

Network science, graph methods, and advanced analytics for problems that are relational — logistics, influence, and multi-actor systems.

Open the practice →

Flagship product

CIPHER

Standard detectors find one intact object. CIPHER finds a threat that has been broken into parts — or split across people and time.

Contraband Interdiction Platform for Hazardous Element Recognition. Built for screening environments where the interesting thing is never the obvious thing.

Hypothetical dual-energy screening scan

Engagement

How a program starts

01 — Assess

Map the mission

Operators, data you actually have, and the decision that has to get better. A written problem statement and a go / no-go in weeks, not quarters.

02 — Engineer

Stand up the working system

A model, a fusion layer, or a decision aid — scoped to the mission and instrumented so you can see where it fails.

03 — Deliver

Hand it to the operators

Integration, review in the loop, and a path to sustainment. If it cannot be used, it is not finished.