We turn the edge into the cloud.
We don’t connect the edge to the cloud. We turn the edge into the cloud. U92 Deeptech is our research programme for sovereign, offline-capable AI and networks in places where connectivity fails.
The U92 Deeptech products
Each product works on its own and gets stronger alongside the others.
KalaMOS
DesignedSovereign edge-AI operating system
KalaMOS runs AI on the device itself, in Light and Standard builds, with no dependence on a remote cloud. Named for Dr APJ Abdul Kalam.
Explore KalaMOSBoSC3
DesignedAgentic command and control
BoSC3 coordinates people and machines in the field and is built to keep working as nodes drop out. Named for J.C. Bose and S.N. Bose.
Explore BoSC3NobisGRID
DesignedSelf-healing mesh network
NobisGRID is a decentralised mesh that reroutes itself when links fail, using stochastic path optimisation. Named for P.C. Mahalanobis.
Explore NobisGRIDMagCHAIN
DesignedTamper-evident custody and provenance
MagCHAIN records who held an asset, where and when, in a ledger that shows any tampering. It also underpins Kayak.
Explore MagCHAINResearch, stated honestly.
Design stage
Every U92 Deeptech product is at design stage. We say so plainly.
Design partners wanted
We are looking for partners in energy, environment and public infrastructure.
Named for Indian science
KalaMOS, BoSC3 and NobisGRID honour Kalam, the Boses and Mahalanobis.
Where U92 Deeptech is used
Healthcare & Pharma
Hospitals, diagnostics and pharma hold sensitive records and run systems that cannot stop.
Read moreAerospace & Defence
Operations that must continue when networks are jammed, cut or denied.
Read moreSatellite & Telecom
Networks that carry everyone else’s data and must be secured for the quantum decade.
Read moreEnergy & Gas
Grids, plants and pipelines spread across remote sites with critical control systems.
Read moreWater & Public Infra
Utilities that communities depend on every hour of the day.
Read moreLogistics & Supply Chain
Goods moving through many hands, where custody and visibility matter.
Read moreBuilding something that cannot fail?
Tell us what you are working on. We will suggest a small, fixed-scope first step.