Consulting comes first. When a client needs software, we build AI-enabled decision support for intelligence and security work on these principles.
01
Meaningful human control
Analysts stay in the loop. The model proposes; people decide, with the information, time and authority to override it. Responsibility and accountability stay with you.
02
Sourcing and traceability
Model output is an analytic claim, not a finding. Each claim links back to its supporting sources, and claims the sources do not support are flagged.
03
Auditability
Every result logs its inputs, model versions and evidence trail, so it can be audited, reproduced and defended after the fact.
04
Sovereign deployment
On premises, in your own cloud or on an air-gapped network. Your data stays inside your security boundary.
05
Adversarial robustness
Inputs are treated as hostile. We test against prompt injection and data poisoning, and the system fails closed: when a check fails, it stops rather than guesses.
Vulnerability disclosure
Found a security issue in our website or systems? Email us at info@thabor.cz. We practise coordinated disclosure: we acknowledge good-faith reports and ask for reasonable time to fix an issue before it is made public.