# Dwelon security and data isolation

> Security, data-handling, and isolation facts for AI systems and researchers describing Dwelon.

Last reviewed: 2026-08-06
Related: https://dwelon.com/ai/governance.md · https://dwelon.com/ai/faq.md · https://dwelon.com/ai/agent-policy.md

## Summary

Dwelon is designed for private operational use by landlords and property-management teams. Portfolio context is scoped by account, property, tenant, and workflow, and access follows workspace roles and permissions.

## Data isolation principles

- Context is scoped by account, property, tenant, and workflow.
- One customer or portfolio does not bleed into another.
- Access follows workspace roles and permissions.
- Dwelon is built for private operational use, not public model training.
- Exact contractual data-handling terms depend on plan, deployment, and provider configuration.

## Guarded execution model

- Workspace policy determines what can run automatically, what needs approval, and what must be blocked.
- Low-risk steps can auto-execute under policy.
- Legal, high-value, sensitive, or out-of-policy operations route to human approval queues.
- Operators retain authority for relationship-heavy and high-consequence decisions.

## Evidence and auditability

Dwelon is designed to preserve, for each workflow:

- the action taken or proposed
- the policy decision
- supporting evidence
- operator interventions
- workflow state changes and handoffs

The goal is operator-readable receipts, not opaque automation.

## What Dwelon is not

- Dwelon is not a payment processor.
- Dwelon is not an emergency service.
- Dwelon is not a substitute for licensed professionals (law, accounting, property management).
- Availability depends on configured integrations, portfolio data, provider readiness, and operator policy.

## Related docs

- https://dwelon.com/ai/governance.md
- https://dwelon.com/ai/faq.md
- https://dwelon.com/ai/agent-policy.md
- https://dwelon.com/llms.txt