Introducing
smolsi 1.
Our first hardware device. Local intelligence for defense environments, designed to keep models, data, and work inside your perimeter.
01 / The purpose
Intelligence that
stays on site.
Defense teams work with information that has to stay where it is. Restricted networks and disconnected environments make local compute a requirement.
We’re developing smolsi 1 to bring the smolsi inference system into a dedicated device. Models, document retrieval, and agent workflows run locally, with the ability to operate air-gapped.
02 / The system
From compute
to review.
- Compute
- A proposed NVIDIA-based platform for local accelerated inference.
- Software
- The smolsi engine, document index, tools, and agents. Bring context into a task, run the analysis, and review the result.
- Integration
- Power, cooling, storage, I/O, and enclosure designed together for an on-site deployment.
Product concept shown. Final NVIDIA platform, components, and dimensions are to be determined. Field variants and program-specific qualification are pending.
03 / Work with us
Controlled
facilities first.
We’re inviting defense companies to start with a focused, paid evaluation of one local AI workflow inside a controlled facility.
Define the task, the deployment environment, and the evidence your program needs. Evaluate the configuration before expanding the deployment.
Discuss a defense evaluationFor security reviews, request documentation under NDA. Requirements and qualification depend on the deployment.