A support bot grounded in your own content.
A floating chat widget that answers visitors from your site content, captures a lead when it is unsure, and logs every conversation. Bring your own Anthropic and OpenAI keys, with no per-conversation markup.
Grounded answers, not guessing
A working bot, fast
An AI onboarding wizard reads your site, interviews you one question at a time, and proposes a full setup. No major competitor ships an interview-to-knowledge setup.
Knowledge base
Hand-written entries, embedded and chunked on save.
Auto-train
Pull from selected posts and pages automatically.
Document import
Import txt, md, csv, docx, and best-effort PDF.
URL import
Import from page or sitemap URLs, same-site guarded.
Onboarding wizard
Connect keys, get interviewed, review, and go live.
Streaming replies
SSE answers with multi-turn history across pages.
Voice in and out
Whisper input and TTS output, with a talking-agent mode.
Vision input
Send an image and the bot reads it with Claude vision.
Lead capture
Captures a lead when confidence is low, with a human handoff.
Conversation flows
A no-code step builder for scripted, branching chats.
Selectable models
Anthropic Claude, OpenAI, or OpenRouter for answers.
Auto-language
Detects each visitor language and replies in it.
Self-hosted, bring your own key, flat license
Flat license, raw model cost
No per-conversation markup. You pay a flat license for the plugin and the raw model spend on your own keys.
What you control
Your Anthropic and OpenAI keys, your knowledge base, your persona and tone, your conversation logs, and your data, all on your own server.
Browse the suite →Get C7 AI Chatbot
The full grounded widget: onboarding wizard, RAG knowledge, voice and vision, conversation flows, lead capture, selectable models, and auto-language.
Get C7 AI Chatbot →Specs
C7 AI Chatbot technical summary.
Platform
WordPress 6.0+.
Answer models
Anthropic Claude, OpenAI, or OpenRouter.
Embeddings and voice
OpenAI, with Whisper in and TTS out.
Knowledge
KB entries, posts, pages, documents, URLs.
Hosting
Self-hosted, bring your own key, flat license.
Privacy
Conversations logged locally; no data leaves the site.
Frequently asked questions
How is the pricing different?
It is self-hosted with a flat license and your own keys, so you pay the raw model cost with no per-conversation or per-resolution markup.
Where do the answers come from?
Only from the knowledge you supply, retrieved over semantic embeddings. A confidence threshold flags uncertainty and offers a human handoff.
How fast can I set it up?
The onboarding wizard reads your site, interviews you one question at a time, and proposes a full setup you review before going live.
Does it support voice?
Yes. Voice input via Whisper and spoken replies via TTS, including an auto-speak talking-agent mode the visitor can mute.
Does visitor data leave my site?
No. Conversations are logged in wp-admin and no visitor data leaves the site beyond the model calls on your own keys.
