How we work with artificial intelligence.
Booka's AI processes sensitive documents of real professionals. This demands transparency, not marketing. Here is our full policy.
Key blocks
Three autonomous agents (Verifier, Regulator, Trust) process, verify and orchestrate international credentialing files. They don't replace humans: they free them from mechanical tasks.
Academic, professional and identity documents. Always with the explicit consent of the holder. Never for purposes other than credentialing.
Enterprise LLM models (Anthropic Claude) with Booka-specific fine-tuning. Auditable training data. No use of personally identifiable data for training.
Continuous auditing. Human-in-the-loop on critical decisions. Traceable logs. Compliance with the European AI Act from day one.
Any candidate or client can request a detailed explanation of how the AI made a decision about their file. Response guaranteed within 72h.
We do not use candidate documents to train models. Agents use Claude via API with zero retention.