Permission-aware AI
Retrieval is filtered by the requesting user's rights before any content reaches the model.
- Pre-retrieval permission filteringAvailable
- Restricted items never disclosed in resultsAvailable
- Windows file rights read by Pulsor Agent and applied before the AI, never wider than WindowsIn progress
Auditability
Visibility into access and administrative actions.
- Audit logsAvailable
- Access visibility — sessions, questions, documents opened and exports per employeeAvailable
- Administrative action historyAvailable
AI security
Controls specific to AI models, prompts and agents.
- Prompt-injection mitigations (policy in every prompt; ongoing work)Partial
- Agent approval workflowsAvailable
- Model and external-source policiesTo confirm
- Multi-model architecture — cloud and private AI modelsAvailable
- Local AI modelsPer deployment
