Public client identifier, server-protected secret, tenant-specific configuration, and domain whitelisting — embedded safely on approved sites only.
A secure widget deployment separates:
The public identifier is not treated as a secret. Sensitive authorization decisions remain server-side.
Each organization can approve the websites or subdomains allowed to initialize its widget. Requests from unapproved origins can be rejected.
Domain restrictions help prevent a copied script from being embedded on another website and consuming the customer’s AI quota or impersonating its support experience.
Rumbe-Widget-V3 is compiled as a lightweight, framework-independent script. Avoiding a runtime dependency on React or another host framework reduces the risk of version conflicts with the customer’s website.
A zero-dependency bundle does not eliminate security review. Content security policy, script integrity, update processes, origin validation, and backend authorization remain important.
With the applicable authenticated-widget capability, a host application can provide a verified user profile to Rumbe. The host should generate signed, short-lived assertions server-side. Sensitive identity data should never be trusted solely because it was supplied by browser JavaScript.
Widget uploads and conversations must remain associated with the active tenant and authenticated or anonymous session. File access URLs should be time-limited or authorization-protected, and unsupported file types should be rejected.
Real-time voice sessions use WebRTC or supported WebSocket connections to configured model providers. Microphone access requires explicit browser permission. Customers should disclose recording and processing practices and apply consent requirements appropriate to their jurisdiction.
The script may be visible, but domain whitelisting and server-side validation are designed to reject unauthorized origins.
No. It identifies the tenant client-side. Privileged validation must rely on protected server-side controls.
An authenticated integration can pass a verified, signed profile when the relevant capability is enabled and securely implemented.
No. It reduces dependency conflicts, but server authorization, origin validation, CSP, updates, and data handling still require review.
Vovance Inc. can discuss Rumbe AI’s architecture, available controls, deployment assumptions, and contractual options for your use case.