Embed Tech Ringer AI's safety-first agent inside your product. JWT auth, per-end-user capability gating, audit logs scoped to your tenant — without building the safety stack yourself.
Four steps. Defaults are conservative — you can broaden later.
Your backend issues a short-lived JWT scoped to one end-user when they open the widget. Tech Ringer AI validates the JWT and creates an agent session for that user.
Add a small JavaScript snippet (or use the React component) to your product. Configure theme, position, and the initial system prompt.
What the agent can do varies per end-user. Capability gating uses claims from your JWT to determine which tools and write actions are allowed.
Every end-user interaction lands in your tenant's audit log. Filter by user, time, and event type to debug or audit.
End-user (in your app): "Help me set up the integration with Stripe."The agent uses connected docs (if read enabled) and walks the user through configuration. Your audit log shows the interaction tied to the JWT-issued end-user identity.
In-app embedding is for teams that want to ship AI inside their own product without rebuilding the safety stack. You probably don't want to maintain a guardrail pipeline, a PII redaction service, a credential isolation layer, and an audit-log infrastructure — that's the entire point of Tech Ringer AI. The embed lets you delegate all of that and surface the agent inside your UX.
The auth model is short-lived JWTs minted by your backend. You stay in control of who can access the agent, and per-end-user claims in the JWT drive capability gating: 'this user can read but not write,' 'this user is a paid tier so the model upgrade is enabled,' 'this user is in a regulated industry so additional content filters apply.' Your backend stays the authority on who-can-do-what.
Audit logs are scoped to your tenant: every interaction is attributable to the JWT-issued end-user identity, which means you can debug a customer's session, prove compliance to a regulator, or revoke a user's access without Tech Ringer AI being in the loop. The same audit log everyone else uses internally, just scoped to per-end-user records.
Two paths: a vanilla JS snippet (`<script src="..." data-tenant="..."></script>` + a JWT-issued from your backend), or a React component (`<TechRingerAgent jwt={...} />`). Both options expose the same configuration surface — theme, position, initial prompt, capability claims.
Yes. The widget accepts theme tokens (colors, font, border radius, position). For deeper customization, you can host your own UI and use the Tech Ringer AI agent SDK directly — the widget is opinionated but optional.
No. We don't train models on customer conversations — across every plan, every audience, every chat platform. PII is also redacted before any third-party LLM provider sees a tool result.
Yes. Input guardrails, URL safety scanning, PII redaction, capability gating, and audit logs run identically here as in the dashboard and on every other integration. Adding a chat platform doesn't expand your security surface area meaningfully.
Disconnect the integration from the dashboard at any time. That revokes the OAuth grant (or removes the bot from the platform, for non-OAuth platforms), all in-flight agent sessions stop, and any future invocations from that platform fail closed. The audit log of past activity stays accessible.
Free to start. $5 in credits on us. No credit card required.