Toulouse, France
SOASY SAS was incorporated in Toulouse in 2023. The company has one purpose: to design, develop and license its own AI software for high-criticality industries. All code and models are built in-house and remain the property of the company.
No institutional or venture funding has been raised to date. Development has been financed by the founders and by product revenue. A first venture round is planned to accelerate product development and commercial deployment.
Revenue comes from software licences on products we own, not from time and materials. The same build ships to every customer; deployment is an installation, not a bespoke project.
Priced per deployment perimeter — per site, per vessel or per fleet — renewed annually.
A fixed-scope, fixed-price run of the product on a sample of the customer's own documentation. The same procedure for every customer.
Maintenance and product updates are included in the licence for its duration.
President & Chief Operating Officer
Electronic systems engineer. Fifteen years across defence, naval, aerospace and rail programmes before founding SOASY.
Institutional & business development
A senior institutional and diplomatic career, applied to market access and international relations.
Domain manager — Naval & defence
Former French Navy officer. Twenty years of operational and command experience, on surface ships and submarines.
The company is founded as a software publisher for critical environments. First engineering work on DESS.AI begins.
DESS.AI and then E-NAUTIS enter qualification with first customers on their own technical documentation.
DESS.AI and E-NAUTIS share one platform doctrine: the AI runs inside the customer's perimeter.
Extending per-site licences to multi-site agreements with existing customers, and first deployments outside France.
Both products served by a single on-premise runtime, so a new sector is a configuration rather than a new codebase.
Our products run inside the customer's perimeter. The engineering that produces them does not. Training, evaluation and build all happen on cloud infrastructure, and only the resulting artefact is packaged for on-premise delivery.
Fine-tuning and retrieval-model training on GPU compute, on synthetic and licensed technical corpora.
Automated regression and retrieval-quality scoring on every build, before a release is cut.
Continuous integration and container packaging of the release that ships to on-premise and embedded targets.
A cloud-hosted instance for demonstrations and for prospects who want to evaluate before an on-premise install.
Customer production data never transits this infrastructure. It stays where the licence installs the product.
Product walkthrough, licensing terms or partnership — the fastest route is a direct message to the team in Toulouse.
