Security by design
Permissions, identity and audit are part of the platform foundation, shared by every module.
- Shared permission model across modulesAvailable
- Least-privilege defaultsTo confirm
Security
Pulsor is built on one principle: AI only sees what the user is authorized to see. This page describes our security approach. Controls marked “To confirm” are being verified and will be published on the Trust Center.
Security
Permissions are enforced before retrieval, not after generation. Two people can ask the same question and receive different, correct answers.
Same questionWhat is the status of Project Atlas?
Project member
3 sources usedAtlas is on schedule for phase 2. Budget consumption is 64% [1]; the vendor review closes on the 18th [2].
External contractor
1 source usedI can share the public project brief [1]. Budget and vendor details are outside your access.
Permissions, identity and audit are part of the platform foundation, shared by every module.
Federated identity and role-based access control.
Retrieval is filtered by the requesting user's rights before any content reaches the model.
To answer a question, the AI model receives the question and the most relevant excerpts the person may read — not your document base. What leaves your installation, and to whom, depends on the deployment you choose.
Protecting customer data in transit, at rest and between tenants.
Visibility into access and administrative actions.
Secure development practices across the platform.
Hardening and operations of the environments Pulsor runs.
Backup and disaster-recovery objectives for managed environments.
How security incidents are detected, handled and communicated.
Controls specific to AI models, prompts and agents.
Private Cloud, On-Premise and Sovereign deployments for maximum control.
A demo shaped around your sources, your security requirements and your deployment model.