Embeddable chat
Drop one script tag. Give every visitor a conversation that carries context.
Navigic's embeddable chat widget adds AI-powered conversations to any website with Shadow DOM isolation, streaming responses, session persistence, and full programmatic control.
ContextDelegationReviewMemory
AudienceProduct and growth teams adding AI-powered chat to their website
The workflow keeps the active context, delegated work, review points, and durable memory visible before the next action moves.
Recognize the friction
Work should not reset between handoffs.
Building a custom chat widget means maintaining Shadow DOM, WebSocket connections, streaming, and session state from scratch.
Bolted-on chatbots feel generic and disconnected from the actual product, eroding trust instead of building it.
Workflow
From first context to reviewed follow-through.
- Embed with one script tag
Add a single script tag to any page. The widget mounts in an isolated Shadow DOM that will not conflict with existing styles or scripts.
- Configure appearance and behavior
Set the brand color, greeting message, title, and position. Use the live demo to preview changes before deploying.
- Connect to your workspace
Point the widget at your Navigic workspace to route conversations through the same context, memory, and review system your team already uses.
- Review conversations and outcomes
Inspect visitor conversations, surface recurring questions, and feed signals back into product and support workflows.
Capabilities
The same operating pattern, adapted to this work.
Shadow DOM isolation
The widget runs inside an isolated Shadow DOM with constructable stylesheets, so host page CSS never leaks in and widget styles never leak out.
Streaming responses
Responses arrive word by word with a typing indicator, matching the cadence users expect from modern AI chat interfaces.
Session persistence
Conversations persist across page navigations and return visits through lightweight localStorage session management.
Programmatic API
Open, close, initialize, and destroy the widget from your own JavaScript. Use NavigicWidget.init(), .open(), .close(), and .destroy() for full control.
Mock mode for testing
Set data-mode="mock" to use canned responses with simulated streaming, so you can develop and test without a live backend.
Safeguards and trust
Keep control visible where the work carries risk.
Navigic should help prepare, preserve, and review work without hiding the human-owned decision points.
Human escalation path
Configure when and how conversations should escalate to a human teammate instead of continuing with AI alone.
Conversation review
Every visitor conversation is available for team review, so automated responses remain visible and correctable.
No sensitive data in storage
Only a session identifier is stored in localStorage. Conversation content stays server-side and under your workspace's data controls.
Configurable boundaries
Set the greeting, topic scope, and response style to keep the widget aligned with what your team is prepared to support.
Related resources
Move from solution shape to practical next steps.
Use cases
Review concrete work patterns for continuity, delegated action, and human control.
Open Use casesTutorials
Start with practical setup paths for memory, review, and visible AI work.
Open TutorialsArgo for Mac
See the current Mac app for carrying active context through one work loop.
Open Argo for MacDownload
Install Argo for Mac when you are ready to try the workflow directly.
Open DownloadWidget demo
Try the embeddable chat widget with a live configurator and instant code snippet.
Open Widget demoBegin
Bring one active loop aboard.
Start with the account path, inspect the Mac app, or download Argo when you are ready to try the workflow.