Workspace isolation
Workspace data is scoped by tenant throughout Alliance, with private channels enforcing access at the communication layer as well.
Security at Alliance
Trust is a product feature, not a promise in the footer.
Alliance is designed so teams can understand what their AI staff knows, what it can do, and what it has done.
Approval by design
External actions can be held for human approval according to workspace policy instead of relying on an agent to ask voluntarily.
Visible accountability
Agent actions produce a record of who acted, what happened, and whether a person or policy approved it.
Controlled credentials
Provider and integration credentials stay in server-side runtime boundaries and are never exposed to browser clients.
AI provider boundaries
The model is part of the system, not the system of record.
Alliance may route work to configured AI model providers. Only the context needed to complete that work is sent to the selected provider; workspace records, policies, and the durable work ledger remain controlled by Alliance.
Available providers and processing terms can vary by workspace configuration. Current providers are disclosed on the subprocessors page.
Report a concern
Help us protect Alliance.
Send suspected vulnerabilities privately to [email protected]. Please include enough detail for us to reproduce and investigate the issue.
Bring useful AI into the business without losing oversight.
Join Alliance in private early access and help shape a better way to work with AI.