Security
Control where work, data, and AI meet.
Start with the boundaries: where BaseState runs, what it connects to, who may see and act on the work, and how consequential changes are reviewed.
Deployment boundary
Start with where the system runs.
A security conversation begins with the deployment, connected systems, customer-owned data, and responsibilities on each side.
- Instance and environment boundary
- Connected-system boundary
- Operational ownership and review
Identity and access
Authority follows the requester and the work.
Access is evaluated in the context of the person, system, or controlled AI action requesting the view or change.
- Who may see the view
- Which actions are available
- Which actions require approval
Change attribution
The record includes who or what acted.
Human decisions, connected-system observations, and AI contributions belong in the same governed history.
- Actor and time remain visible
- Evidence stays connected
- Review does not depend on narration
AI boundaries
The apprentice has no back channel.
AI drafts and acts through the same available actions, approval paths, and record as people. Consequential work remains explicit.
- No invisible write path
- Approval remains part of the command
- Outcome returns to the record
Evaluation posture
Claims follow evidence.
Deployment, encryption, retention, standards, and evidence-handling answers are scoped to the actual implementation and reviewed in the security conversation.
Talk to the founders.
Tell us how your team works and what gets in the way. We’ll look at the process together and see where BaseState can help.
45-minute video call with Paul or Hanyi.