Got Questions?
We've Got Answers

Wisdone AI is an AI-powered chatbot widget that you can embed on your website. It uses RAG (Retrieval-Augmented Generation) technology to answer visitor questions accurately based on your own content—your website pages, documents, and FAQs. Unlike generic chatbots, Wisdone learns your specific content and provides answers with source citations.

You can have your AI chatbot live on your website in under 5 minutes. Simply create an account, add your website URL or upload documents, customize the appearance, and paste the embed code on your site. Wisdone automatically crawls and indexes your content.

After configuring your chatbot in the dashboard, you'll receive a simple one-line script tag. Just paste this code before the closing </body> tag on your website. It works with any website platform—WordPress, Shopify, Wix, Squarespace, custom HTML, or any CMS. No coding knowledge required.

Yes. Deploying the same bot across several domains you own is included on the Pro plan and above (multi-domain support). Separately, the number of website sources you can crawl into your knowledge base depends on your plan: Starter includes 3 websites, Pro includes 10, and Enterprise is unlimited. Each source keeps its own crawled content.

RAG (Retrieval-Augmented Generation) is an AI technique that combines semantic search with language models. Here's the flow: 1) Your content is processed into vector embeddings and stored in a vector store, 2) When a visitor asks a question, it's converted to a vector, 3) The vector store finds the most semantically similar content chunks, 4) These chunks are sent to the LLM as context, 5) The LLM generates a response grounded in your actual content. This means configuring RAG = configuring your knowledge base. Better content = better answers.

Every response from your AI chatbot can include clickable citations that link back to the original source—whether it's a page on your website or an uploaded document. This builds trust with visitors by showing exactly where the information comes from. You can enable or disable citations in your dashboard settings.

Absolutely! You have full control over the chatbot's look and feel. Customize primary and secondary colors, choose between light/dark/auto themes, enable glassmorphism effects, upload your own logo, set custom greetings, configure suggested questions, and adjust the widget button size and animation. Everything updates in real-time.

Yes! You can configure the AI's persona with a custom name and description, set the response style (concise, detailed, or conversational), adjust the temperature for creativity, and add custom system prompts for specific instructions. On the Pro plan and above you can also choose between OpenAI and Anthropic (Claude) models.

Smart CTAs (Call-to-Action buttons) appear contextually during conversations. You can configure buttons that trigger when visitors show buying intent—like "Book a Demo", "Get Pricing", or "Contact Sales". CTAs can open URLs, send emails, make phone calls, trigger webhooks, or send pre-filled messages. The AI decides when to show them based on conversation context.

Yes! The underlying AI models support over 100 languages. Your chatbot will respond in the same language the visitor uses. For best results, include content in the languages you want to support in your knowledge base.

Yes. When a visitor asks for a person, or the AI repeatedly can't resolve a request, Wisdone can escalate the conversation to your team. You can receive escalations by email, in a Slack channel (where your team replies directly and messages relay back to the visitor in real time), or via a webhook to your own system. You can also set business hours so after-hours conversations are handled differently.

Your dashboard shows what visitors are actually asking—top questions, conversation volume, and content gaps where the bot had no good answer—plus visitor satisfaction ratings so you can see how helpful it is. Use these insights to improve your knowledge base over time. History retention depends on your plan: 30 days on Free, 90 on Starter, 180 on Pro, and unlimited on Enterprise.

Yes, on the Pro plan and above. Pro includes 5 seats (you plus 4 teammates) and Enterprise is unlimited; Free and Starter are single-user. Invite teammates from your dashboard so they can help manage the knowledge base, review leads, and handle conversations under one shared account.

Configuring RAG (Retrieval-Augmented Generation) is simply configuring your knowledge base. When you add content—whether through website crawling, file uploads, or direct text input—Wisdone automatically processes it through the RAG pipeline: extracting text, splitting it into chunks, generating vector embeddings, and storing them in a vector store for semantic search. There's no separate "RAG configuration"—your knowledge base IS your RAG configuration.

A vector store is a specialized database that stores your content as mathematical embeddings—numerical representations that capture semantic meaning. This enables the AI to find relevant content based on meaning, not just keywords. You don't need to configure it directly; Wisdone handles all vector store operations automatically when you manage your knowledge base. Each account gets an isolated vector collection for security.

When you add content, Wisdone runs it through a multi-step pipeline: 1) Content extraction (parsing HTML, PDFs, docs), 2) Text chunking (splitting into ~1000-character segments with overlap), 3) Embedding generation (converting text to vectors), 4) Vector storage (saving to the vector store with metadata). When visitors ask questions, the same embedding process converts their query to a vector, then the vector store finds the most semantically similar chunks to include as context for the AI response.

In your dashboard, go to Knowledge Base → Websites and enter your URL. You can crawl a single page, enable recursive crawling (up to 2 levels deep), or provide a sitemap URL for large sites. Wisdone automatically extracts text, generates vector embeddings, and indexes everything in your RAG pipeline. The indexed content becomes immediately available for AI-powered question answering.

Wisdone supports PDF documents, Word files (.docx), plain text files (.txt), CSV spreadsheets, and JSON files. Simply drag and drop files in the Knowledge Base → Files section. Content is automatically extracted, chunked into optimal segments, converted to vector embeddings, and indexed in the vector store for semantic search. Each file becomes part of your RAG knowledge base.

You can configure automatic content synchronization on a daily or weekly schedule. Wisdone uses content hashing to detect changes—only modified pages are re-indexed, saving processing time and embedding costs. When content changes, the old vectors are replaced with new ones in your vector store. You can also trigger a manual sync anytime from your dashboard.

Content limits depend on your plan and are measured by the number of sources you add: - Starter: 3 websites (up to 500 pages each), 20 files, 50 text snippets, - Pro: 10 websites (up to 2,000 pages each), 200 files, 300 text snippets, - Enterprise: unlimited sources, up to 10,000 pages each. Each website source crawls its pages automatically, following links or reading your sitemap, so a single source usually covers your whole site. The vector store keeps semantic search fast no matter how much content you add.

Answer accuracy depends on your knowledge base quality. Tips: 1) Add comprehensive content covering topics visitors ask about, 2) Include FAQ-style content with clear questions and answers, 3) Remove outdated or contradictory information, 4) Use the retrieval depth setting (1-20 documents) to control how much context the AI receives, 5) Write clear, well-structured content—the AI works best with organized information. Better knowledge base content = better RAG performance = more accurate answers.

Wisdone captures leads naturally during conversations. When visitors ask about pricing, request demos, or show buying intent, the AI prompts them for contact information. You can collect email, phone, name, or custom fields. Leads are stored in your dashboard and can trigger instant email notifications. Lead capture is available on the Starter plan and above.

Yes! You can set the capture mode to Passive (only when users volunteer info), Proactive (suggest capture after trigger conditions), or Aggressive (actively prompt for information). You can also configure triggers like "after N questions" or on specific keywords like "pricing" or "demo".

Enable email notifications in your dashboard under Notifications. You can add multiple email addresses to receive instant alerts when leads are captured. Each notification includes the visitor's contact info, conversation summary, and the question that triggered the capture.

Yes! All captured leads are accessible in your dashboard under the Leads section. You can view conversation history, contact details, and metadata like IP address and user agent. Export functionality allows you to download leads for import into your CRM or email marketing tools.

Wisdone connects with the tools you already use: - Slack — hand live conversations off to your team and reply straight from a Slack channel, - Webhooks — trigger Zapier, Make, your CRM, or any custom API from CTA buttons or when a chat is escalated, - Email — get instant notifications when new leads come in or a conversation is escalated, - Any website platform — the widget embeds on WordPress, Shopify, Wix, Squarespace, Webflow, or custom HTML with a single line of code. Lead data can also be exported anytime for manual import into other systems.

Yes. The widget works on any Shopify store: in your Shopify admin, go to Online Store → Themes → Edit code, open theme.liquid, and paste the script tag just before the closing </body> tag. The assistant then answers from your store's pages, like any other website.

Yes. When the chatbot escalates a conversation—because a visitor asks for a human, or the AI can't resolve the request—it can post the conversation to a Slack channel you choose. Your team replies right from Slack, and those messages are relayed back to the visitor in the chat in real time. No one has to sit inside a separate dashboard waiting for handoffs.

Yes, in two places. First, CTA buttons can fire a webhook when clicked—use it to kick off a Zapier or Make workflow, push data to your CRM, or call your own API. Second, escalations can be sent to a webhook endpoint with the full conversation transcript and visitor details, secured with an optional signing secret and custom headers. Webhooks are the most flexible way to connect Wisdone to almost any external system.

Yes—today this works through webhooks and exports. Point a CTA or escalation webhook at Zapier or Make to forward captured leads into your CRM, email marketing platform, or a spreadsheet automatically. You can also export leads directly from the dashboard for manual import. Native one-click CRM connectors are on our roadmap.

There are four tiers. Free gives you 50 messages a month on one website, with no credit card and no time limit. Starter ($19/month) includes 1,000 messages and 3 websites, Pro ($59/month) includes 10,000 messages and 10 websites, and Enterprise ($199/month) is unlimited on both. Pricing is flat — there is no per-answer or per-resolution fee, so the bill does not grow when the chatbot does its job well. Every tier includes RAG-powered answers, source citations, lead capture, and full customization; Pro adds multi-domain support, team members, and Anthropic (Claude) models.

Each message sent by a visitor counts toward your monthly limit. AI responses don't count against your limit. When you reach your limit, the chatbot gracefully informs visitors and can display a fallback contact method. Unused messages don't roll over to the next month, but if you need more you can top up with a message pack (500 extra messages) without changing your plan.

Yes! You can change your plan anytime from the Billing section in your dashboard. Upgrades take effect immediately with prorated billing, and downgrades take effect at the start of your next billing cycle. You can also cancel whenever you like—your plan stays active until the end of the current period.

Yes! You can start on the free plan—50 messages per month, no credit card required—and use it for as long as you like to try the core experience. Lead capture, integrations, and higher limits unlock when you upgrade to a paid plan, which you can do anytime.

Security is a top priority. Your data is isolated in separate database collections and vector stores—each account gets its own isolated collection. We use JWT-based authentication, HTTPS encryption, and never share your data with third parties. Your content embeddings are stored separately from other users. We're GDPR-compliant and never use your content to train AI models.

Yes. You can export your account data—including your knowledge sources, leads, and settings—from the dashboard at any time. If you choose to delete your account, we permanently remove all of your data, including your indexed content and vector embeddings. Visitor conversations respect your privacy policy, and end users can be shown a consent notice in the chat widget.

Response times typically range from 1-3 seconds depending on query complexity and the AI model selected. The RAG pipeline is optimized for speed: vector search returns relevant documents in milliseconds, then the LLM generates the response. We stream responses in real-time so visitors see the answer being generated.

Wisdone uses OpenAI's GPT models by default. On the Pro plan and above you can switch to Anthropic's Claude models instead, and on any plan you can adjust the response temperature to tune how precise or creative the answers are.

RAG (Retrieval-Augmented Generation) offers several advantages over fine-tuning for knowledge-based chatbots: 1) Instant updates—add new content and it's immediately available, no retraining needed, 2) Source citations—RAG can show exactly where answers come from, 3) Cost-effective—no expensive model training, 4) Accuracy—answers are grounded in your actual content, reducing hallucinations, 5) Privacy—your content stays in the vector store, not baked into model weights.

Embeddings are numerical representations (vectors) of text that capture semantic meaning. When you add content, Wisdone converts each text chunk into a vector. Similar concepts end up with similar vectors. When a visitor asks a question, their query is also converted to a vector, and the vector store finds the content vectors closest in meaning (not just keywords). This enables understanding questions like "return policy" even if your content says "refund guidelines".

Absolutely! The widget is fully responsive and optimized for mobile devices. On smaller screens, it appears as a floating button that opens a bottom sheet chat interface. Touch interactions, keyboard handling, and safe area insets are all properly handled.

Currently, Wisdone manages all infrastructure including the vector store, ensuring optimal performance and security. Enterprise customers with specific compliance requirements can discuss dedicated infrastructure options. Contact us for custom deployment needs.