We build AI for defense organizations that has to work in the field, not just in the demonstration. Autonomous systems, sensor fusion, intelligence workflows, and the secure deployment infrastructure underneath them, as a prime where scope allows, or as the AI capability inside a prime's team.
Defense AI has no shortage of prototypes. What is scarce is the engineering that takes a promising model and makes it something an operator can rely on: bounded latency, understood failure modes, degraded-mode behaviour, a deployment path into a classified environment, and documentation that survives a change of personnel.
That is the work we do. We won the NATO Defence Innovation Challenge in 2020 with Reticle, and have delivered for the Department of National Defence and the Canadian Armed Forces, U.S. federal government clients, NATO allies and defense primes, and Innovation, Science and Economic Development Canada.
Computer vision and reinforcement learning for unmanned platforms: counter-UAS classification distinguishing drones from birds and aircraft by behavioural signature, drone swarm logic, and radar sensor fusion.
NLP and large language models applied to signal and document analysis: OSINT discovery tooling, counterterrorism intelligence workflows, and automated document triage.
Real-time radio spectrum analysis and anomaly detection, so operators are alerted to a developing situation rather than reconstructing it afterwards.
Fleet readiness modelling from telemetry and maintenance records, targeted at availability rather than dashboards.
On-prem and air-gapped infrastructure, ISO 42001-aligned governance, and risk mitigation for AI operating in contested environments.
AI-generated scenarios for evaluating detection and response under conditions that cannot be staged physically. See AI Research.
U.S. and NATO registration is already in place, so teaming does not wait on paperwork. Every identifier is verifiable in a public registry.
Registry entries establish procurement eligibility. They do not imply endorsement by any government. Full detail is in our capability statement.
Both, depending on scope. We are comfortable as the AI capability inside a prime's team, and we hold the registrations needed to be added to one. For larger integrated programs, subcontracting is usually the honest answer.
Members of our team hold Top Secret clearance, and we are registered under Canada's Controlled Goods Program. Deployment into disconnected and air-gapped environments is standard practice for us, not an exception.
For custom engagements, clients own project-specific code, models, and documentation as defined in the applicable agreement. Where a pre-existing Lemay.ai platform is selected it is licensed separately, and that licence does not reach what you build on top of it.
A 30-minute discovery call, then a fixed-fee scope with deliverables, milestones, and price, typically within two weeks. Defense and government enquiries go to [email protected].
Tell us the program and the constraint. If there is an export-control or clearance dimension, say so early and we will scope it properly.