內容簡介
Inqyra 是一款智能聊天機器人外掛，能根據您的網站頁面、文章及文件回答訪客問題。使用者可連接自己的 AI 提供者，確保數據安全並僅支付實際使用的成本。
【主要功能】
• 支援 5 種 AI 提供者：OpenAI、Gemini、Claude、Mistral、DeepSeek
• 瀏覽器端嵌入：內容不會離開伺服器進行索引
• 自動重新索引：當更新文章時自動重新索引內容
• 自訂小工具：可調整顏色、位置及問候語
• 成本與令牌追蹤：監控 API 使用情況並設定預算限制
外掛標籤
開發者團隊
原文外掛簡介
Inqyra adds an intelligent chatbot to your WordPress site that answers visitor questions based on your pages, posts, and documents. Connect your own AI provider, keep your data on your server, and pay only what the AI actually costs.
Bring Your Own Key (BYOK) — You connect directly to your AI provider. No middleman, no markup. A typical conversation costs ~$0.003 with GPT-4o-mini.
5 AI Providers
Choose from OpenAI, Google Gemini, Anthropic (Claude), Mistral, or DeepSeek. Switch providers or models at any time — your content stays indexed.
How It Works
Inqyra uses Retrieval-Augmented Generation (RAG) to ground every answer in your actual content:
Crawl — Inqyra indexes your pages, posts, and custom post types
Embed — Content is converted to vector embeddings entirely in your browser (your content never leaves your server)
Search — When a visitor asks a question, Inqyra finds the most relevant content using vector similarity, keyword matching, and title/path boosting
Answer — The relevant context is sent to your chosen AI provider, which generates a grounded response
Key Features
5 AI providers — OpenAI, Gemini, Claude, Mistral, DeepSeek
Browser-based embeddings — Content never leaves your server for indexing
29 languages — Dutch, English, German, French, Spanish, Italian, Portuguese, and 22 more
Customizable widget — Colors, position, greeting, header, powered-by text
Auto-recrawl — Content re-indexes automatically when you update a post
Cost & token tracking — Monitor API usage and set budget limits
Rate limiting — Built-in IP-based rate limiting (20/min, 200/hr)
API key encryption — Keys encrypted at rest with AES-256-CBC
No cookies — The plugin does not set any cookies
Lightweight — Loads asynchronously with minimal impact on Core Web Vitals
Setup wizard — Live in under 5 minutes with the guided setup
Free Features
Floating chat widget with full customization
Unlimited pages indexed (post type: pages)
All 5 AI providers (BYOK)
Browser-based embeddings
RAG-powered answers
Conversation tracking (last 30 days)
Auto-recrawl on content changes
Cost & token tracking with budget limits
Query rewriting (LLM-powered query expansion)
29-language support
Premium Features
Index posts, products & custom post types
Document upload (PDF, DOCX)
Synonym management for improved search matching
Webhooks (Zapier, Make, CRM integration)
Lead extraction (email, phone, name, company)
Advanced analytics & statistics export
Page targeting (show widget on specific pages)
Extended conversation history & analytics
Escalation & appointment request detection
Email support
Works With Any Theme
Compatible with all modern WordPress themes and page builders including Elementor, Divi, Gutenberg, and WPBakery.
Privacy First
Embeddings are generated entirely in the browser — your content never leaves your server during indexing
Only when a visitor asks a question does relevant context go to the AI provider, through your own API key
No cookies are set by the plugin
GDPR-friendly architecture
Built in the Netherlands by Ipportunities B.V.
External services
This plugin connects to external third-party services to provide AI chatbot functionality. All connections require the site administrator to explicitly configure an API key before any data is sent. No data is transmitted without the administrator’s action.
AI Chat Providers
When a visitor sends a message through the chat widget, the visitor’s message along with relevant content from your website (retrieved via RAG search) and the system prompt configured by the administrator are sent to the chosen AI provider. The administrator selects which provider to use and provides their own API key. Only one provider is active at a time.
Anthropic (Claude)
API endpoint: https://api.anthropic.com/v1/messages
Used for: Generating chat responses based on your website content.
Terms of Service | Privacy Policy
OpenAI
API endpoint: https://api.openai.com/v1/chat/completions
Used for: Generating chat responses based on your website content.
Terms of Use | Privacy Policy
Google Gemini
API endpoint: https://generativelanguage.googleapis.com/v1beta/models/
Used for: Generating chat responses based on your website content.
Terms of Service | Privacy Policy
Mistral AI
API endpoint: https://api.mistral.ai/v1/chat/completions
Used for: Generating chat responses based on your website content.
Terms of Service | Privacy Policy
DeepSeek
API endpoint: https://api.deepseek.com/v1/chat/completions
Used for: Generating chat responses based on your website content.
Terms of Service | Privacy Policy
HuggingFace Transformers.js (ML Inference Service)
This plugin uses the Transformers.js library to perform machine learning inference entirely in the browser for generating vector embeddings. The JavaScript library itself (~850 KB) is bundled locally within the plugin at assets/js/vendor/transformers.min.js.
When the administrator triggers content indexing, the library downloads the ML model files (~50 MB) and WebAssembly runtime from HuggingFace CDN to the user’s browser. This is a service — the model files cannot be bundled locally in the plugin due to their size.
No website content or user data is sent to HuggingFace — the model runs locally in the browser after download. The CDN only serves the model files.
Library: @huggingface/transformers v3.4.1 (bundled locally)
Model: Xenova/paraphrase-multilingual-MiniLM-L12-v2 (loaded from huggingface.co at runtime)
HuggingFace Terms | HuggingFace Privacy
Freemius
This plugin uses Freemius for license management, plugin updates, and optional usage analytics. When the plugin is activated, Freemius may communicate with its servers to validate licenses, deliver updates, and collect anonymous usage data (if the user opts in). Users are prompted for consent before any analytics data is collected.
Terms of Service | Privacy Policy
