Clanker Support — AI Chat & Customer Support Widget
Clanker Support puts an AI support agent on your WordPress site. Visitors get streaming answers sourced from your knowledge base, can escalate to a human when the AI isn’t enough, and replies you send from the dashboard inbox (or by email) appear in the widget live.
Features
- One-minute setup — install, paste your project key under Settings → Clanker Support, done. The launcher bubble appears on every page.
- Answers from your knowledge base — feed it docs URLs, text snippets, and hand-written Q&A in the dashboard; the AI answers with that context.
- Human handoff — configure how many messages a visitor sends before “Talk to a human” appears; escalations notify you by email and Slack.
- Live operator replies — answer from the dashboard inbox or simply reply to the notification email; either way the visitor sees it in the widget.
- Floating bubble or inline — use the site-wide bubble, the
[clanker_support]shortcode to embed the chat inside a page (e.g. Contact), or both. - Brand color — match the launcher and chat bubbles to your site, straight from the settings page.
- Connection check — the settings page verifies your project key against the API and tells you when something is misconfigured.
- Open source & self-hostable — the whole platform is open source (theopenco/llmchat); point the API URL setting at your own deployment if you self-host.
External services
This plugin is a connector for the Clanker Support service. When enabled, it loads the widget script from your configured Clanker Support API origin (https://api.clankersupport.com by default) on your site’s public pages, and messages that visitors type into the widget are sent to that service to generate answers and manage conversations. The settings page additionally makes a server-side request to the same API origin to verify your project key. No visitor data is sent until a visitor interacts with the widget.
- Service: Clanker Support
- Terms of use
- Privacy policy
If you self-host Clanker Support, all of the above requests go to your own deployment instead.