Miguel Sanchez
Fifteen-plus years of production engineering, the recent ones building AI platforms for environments where evidence, auditability and restraint are requirements rather than preferences.
LinkedIn profileCompany
Built by engineers who have operated sensitive systems in production
Help organisations with real secrets use AI on their own terms, with information crossings that are protected, controlled, and provable.
Why Kordane exists
- Sensitive AI workflows are either blocked outright or routed through platforms built for breadth rather than proof.
- Sovereign hosting answers where AI runs. It does not by itself prove why a particular execution was allowed.
- Large models are often used without evidence that the task required them.
- Audit trails frequently record that something happened without retaining the basis for the decision.
That work shaped a conviction: organisations with sensitive information do not need more AI capability. They need control over the crossings. What leaves, what may be seen, and what must be provable afterwards.
Founders
Carlos Merchán
Fifteen-plus years across high-volume production services in banking, betting and conversational AI.
LinkedIn profileKordane is being developed from Dubai and Andalusia, Spain, with an international outlook.
Principles
- Control before convenience. A crossing that cannot be governed does not happen.
- Evidence before assertion. Capability claims carry a status; demonstrations are labelled synthetic.
- Explicit limitations. The security page states what the platform does not solve alone.
- Smaller models where sufficient. Frontier models are for tasks that genuinely require them.
- Humans stay responsible. Consequential actions can require a named person to approve.
Where the company is today
Kordane is currently in private development, consolidating working product systems under one coherent platform. The public demonstration is synthetic, and the pilot programme turns existing capability into customer-specific deployment against real workflows, environments, and security requirements.