Trust & compliance
We sell decision-grade trust: compliance is not an add-on, it is in the code.
Copyright-safe
We do not republish third-party content: we keep metadata, numbers, canonical links. For paid sources we show our own short, transformative summary, with a similarity filter and a disclaimer.
GDPR
Record of processing activities, legal basis, DPIA for thematic profiling; data export and deletion on request (privacy@antesigna.com, response within 30 days). Encrypted backups, configured retention. No surveillance of natural persons.
No financial advice
Theme-only analysis: no personalized recommendation on instruments, no handling of any individual's portfolio. Automatic disclaimers on every financial output.
Ethics & dual-use
Hard filters block targeting/surveillance of individuals, military use cases and personalized advice. Actor models cover only states and public roles.
Provability
Every output carries its evidence chain โ sources traced back to the original, confidence class and factual anchor โ and an immutable audit log. The numbers are computed by code; the LLM writes only the prose, with grounding and an honest refusal when the evidence is not enough.
Security
Secrets outside the code, TLS, strict CORS, input validation, OWASP top-10. Passwords hashed with bcrypt (cost 12), JWT sessions, RBAC and multi-tenant isolation.