AnswerRails adds an AI-powered chat widget to a WordPress site so visitors can get answers drawn from the site’s own content instead of generic responses. It suits site owners who want a customer-facing assistant that stays on topic, cites where its answers come from, and admits when it does not know something rather than inventing an answer.
The assistant works with an AI provider of the site owner’s choosing: OpenAI, Mistral, or Anthropic Claude, using any model in that provider’s catalogue.
A personal API key from the chosen provider is required, and usage is billed to that provider account rather than through a middleman; new installs default to Mistral, which offers a free tier useful for trying the plugin out. Answers, citations, chat logs, and system health information all appear within the WordPress dashboard, under the plugin’s own menu.
Powered by Fabio ChatBot
Is AnswerRails yours?
Tell readers what is limited and unlimited in the free version, and what the AI does in your plugin. Help them decide faster and reach the people already looking at it. In return, you answer their questions.
Key Features of AnswerRails
Chat Widget And Accessibility
Customizable front-end widget. The chat widget can be themed to match a site’s branding, with configurable title, subtitle, avatar, and welcome message, plus a floating launcher. It supports automatic dark mode, is draggable and resizable on desktop, and switches to a full-screen view on phones.
Built for accessibility. The widget is built to WCAG 2.2 AA standards, with screen-reader announcements, full keyboard operation, visible focus states, reduced-motion support, and an admin contrast guard so brand colors stay readable.
Grounded, Honest Answers
Answers grounded in your content. Rather than relying on generic model knowledge, the assistant draws answers from pages, posts, documents, and JSON files added to its knowledge base. Content sourced from posts, pages, or other tracked items stays current automatically, and any item can also be pinned to a manually approved version.
Source citations on every answer. Grounded answers include a visible list of the sources used, giving visitors and site owners a way to check where an answer came from.
Declines rather than invents. When the knowledge base cannot support an answer, the assistant says so instead of guessing. A built-in check also blocks any answer that would link to a page that does not actually exist on the site.
Accurate internal links. The assistant keeps track of a site’s real published pages so it only links to pages that exist. A mistyped link is corrected automatically when one clear match is found, and links that misrepresent their target are removed before the answer is shown.
Cross-reference lookups. A JSON knowledge item can be marked as a cross-reference table, such as mapping a product code to a product page. When a visitor’s question matches a listed term, the assistant automatically runs a follow-up search on the mapped item before answering.
Reliability And Visibility
Fallback during outages. The knowledge base is also indexed locally in the site’s own WordPress database, so if a cloud knowledge-search service goes down, the assistant keeps answering from that local copy instead of failing.
Health dashboard. An overview screen shows which model is answering, whether knowledge search is functioning, and the status of each stage of the pipeline, with a one-click test of the services the assistant depends on.
Chat logs and activity overview. Every conversation is logged with the pipeline path that produced each reply, including a record of what was shown to visitors even when a provider outage caused a failed response.
Privacy And Protection
Full privacy and consent controls. Conversation logging can be set to off, anonymized, or identified, with a configurable retention period, consent prompts recorded as server-side receipts, and visitor control over what the chat stores in their browser. GDPR export, anonymization, and erasure tools are built into a Privacy panel on the chat logs screen, alongside WordPress’s own personal data tools.
Abuse and spam protection. Every visitor message is screened against the connected provider’s content-safety service before it reaches the chat model, with no way to turn this off. Optional Google reCAPTCHA and built-in rate limiting add further protection against spam on the chat endpoint.
How to Install AnswerRails
Step 1: Install the plugin. In the WordPress dashboard, go to Plugins > Add New Plugin, search for AnswerRails, click Install Now, then Activate.
Step 2: Connect an AI provider. Go to AnswerRails > Settings, enter an API key from OpenAI, Mistral, or Anthropic, and choose a model. New installs default to Mistral, which offers a free tier for evaluation.
Step 3: Move to a paid footing for live use. Before running the assistant on a live site, put the chosen provider account on a paid footing, such as pay-as-you-go on Mistral or available credit on OpenAI or Anthropic, since free-tier usage has no reserved capacity.
Step 4: Build the knowledge base. Use the content picker to add pages, posts, documents, or JSON files to the knowledge base so answers are grounded in the site’s own material.
Step 5: Configure appearance and privacy. Adjust the widget’s look under its appearance settings, then set logging mode, retention period, and consent options under the privacy settings.
Good to Know About AnswerRails
The developer notes that AnswerRails requires a personal API key from the chosen AI provider, since usage is billed directly to that account rather than bundled with the plugin. Without an OpenAI key, knowledge search runs locally in the WordPress database; adding one allows an optional vector-based search synced to that provider account.
Chat logs, consent receipts, and the knowledge base are stored in the site’s own database. GDPR export and erasure requests are handled through WordPress’s built-in personal data tools together with a dedicated privacy panel on the chat logs screen.
Uninstalling removes API keys by default while keeping content and logs unless full data deletion is explicitly enabled beforehand. The plugin is built for standard single-site WordPress installs and does not currently support Multisite.
Download AnswerRails on WordPress.org