Vitalii Popov

<p>This plugin is an AI chatbot that integrates with ChatGPT via API by OpenAI. After installation and customization, a chat bubble will appear on your website. The plugin will talk to your website visitors like a live person, motivating visitors to take certain actions. This is something like a sales manager, who will help users navigate the site and advise on your products and services.</p>
<p>Our plugin is based on the most advanced AI models from OpenAI. We use the ChatGPT-4o model. Please note that for correct work you need a working account on OpenAI platform site and some funds on your balance.</p>
<p>The plugin has flexible customization, and allows you to define very precisely how it should communicate with your site visitors. You can ask the AI to motivate users to make purchases, visit certain pages, or perform other actions. The AI assistant will try to guide users exactly in such a way that they are maximally active on your site.</p>
<p>The plugin will prompt users with links to various pages on your site. If you sell products through your site, a visitor will be able to ask the plugin to find products based on a given set of criteria. As a response, the AI will be able to return a direct link to specific pages on the site. In other words, you can also think of the plugin as a super-smart search engine for your site that will help the visitor not only find the information they need, but also take the necessary actions on your site.</p>
<h4>Plugin Features:</h4>
<ul>
<li>Communication with visitors in human language ( any languages supported by OpenAI )</li>
<li>Accurate answers based on information from your website</li>
<li>Direct links to pages on your site as answers to questions</li>
<li>Customizable nature of communication. Several unique characters-consultants.</li>
<li>Customizable limits and precise control over API usage</li>
<li>Detailed logging of dialogs with users</li>
<li>Customizable level of humor. A robot that can joke!</li>
<li>Custom instructions on how to communicate with users</li>
<li>Flexible system of training and re-training of AI assistant</li>
</ul>
<h3>External services</h3>
<p>For its operation, our plugin uses the API from the OpenAI platform. This is vital for our plugin to work because our plugin uses ChatGPT to train the AI assistant and then communicate with users. We pass a minimum of necessary information through the API. We call the following API methods for the following needs:</p>
<ul>
<li>https://api.openai.com/v1/chat/completions (POST) – the call is used to test the plugin by validating your API key and initially saving the settings. Usually this call is made at the moment of initial plugin configuration, when you specify your API key. As part of this call, we send your API key as well as a set of test messages, such as “Say this is a test!”.</li>
<li>https://api.openai.com/v1/files (POST) – the call is used to upload a training file with a list of articles from your site, as well as links to the pages of your file. This file has JSON format. The content of the file is created on the basis of the pages that you decided to use for training the AI assistant when working with the “Training” tab. The file contains the texts of articles from the pages of your site, as well as links to the corresponding pages of your site. A copy of this file is stored on your site, in the plugins/talk-to-website/admin/training directory.  As part of this call, we send the file itself, as well as your API key. This function is called when you perform plugin training on the “Training” tab. We only send information about the articles that you have specified in the training process using the checkboxes.</li>
<li>https://api.openai.com/v1/files/file_id (DELETE) – the call is used to delete an old file that was used in past AI training. As part of this call, we send the ID of the file to be deleted in the file_id parameter, as well as your API key. This function is called when you perform training of the plugin on the “Training” tab, in case we found old files used for past training.</li>
<li>https://api.openai.com/v1/vector_stores (POST) – the call is used when training the plugin, after loading a file using the v1/files method. After uploading a training file, we create a vector store where we put the newly uploaded file. Accordingly, this happens only when training the plugin in the “Training” tab. Within this call, we send the ID of the created file, the name of the vector storage, as well as your API key.</li>
<li>https://api.openai.com/v1/vector_stores/store_id (DELETE) – the call is used to delete the old vector store that was created during the past AI training. This only happens when training the plugin in the “Training” tab. As part of this call, we send the storage ID in the store_id parameter, as well as your API key.</li>
<li>https://api.openai.com/v1/assistants (POST) – the call is used when training the plugin, after creating a vector storage using method v1/vector_stores. Also, this call is used when changing the parameters of the AI assistant (e.g. robot name, additional instructions, company contact phone number, company address, prohibited topics for conversation, and so on – text fields from the “settings” tab inside the plugin). As a result of this call, an AI assistant is created on the OpenAI platform, trained according to the specified settings, and associated with a training file with a list of materials from your site. As part of this call, we pass your API key, as well as various textual information from the “Settings” tab: type of GPT model, robot name, instructions on the robot’s character, general textual instructions for training, forbidden topics and words, humor level, language used to communicate with visitors, contact information about your project (Email, phone number, website address, map address), limits on the length of responses, your website address, a brief description of your website, the ID of the associated vector store.</li>
<li>https://api.openai.com/v1/assistants/assistant_id (DELETE) – the call is used to delete old AI assistants on the OpenAI platform that are no longer needed. As part of this call, we send the assistant ID in the assistant_id parameter, as well as your API key.</li>
<li>https://api.openai.com/v1/threads (POST) – the call is used to create a new conversation thread with the AI assistant. It is used when starting a new dialog between a site visitor and the AI assistant. As part of this call, we send your API key.</li>
<li>https://api.openai.com/v1/threads/thread_id/messages (POST) – the call is used to create a new message from the user to the AI. It is used each time a message is sent to the chat room from the user. The “thread_id” parameter in the URL is the identifier of the thread that was created using the v1/threads call. Within this call, we pass the thread id as well as the text of the message entered by the user in the chat window. Also within this call, we pass your API key.</li>
<li>https://api.openai.com/v1/threads/thread_id/messages (GET) – the call is used to get a list of messages that were sent to the AI or received as a reply from the AI. It is used every time when sending messages to the chat from the user, as well as when receiving a list of responses from the AI. The “thread_id” parameter in the URL is the identifier of the thread that was created using the v1/threads call. Within this call we also pass your API key.</li>
<li>https://api.openai.com/v1/threads/thread_id/runs (POST) – the call is used to create a run within the AI thread. In other words, causes the AI assistant to analyze the messages sent, and generate responses to those messages. It is used every time chat messages are sent from a user. The “thread_id” parameter in the URL is the identifier of the thread that was created by the v1/threads call. Also, within this call we pass your API key, and also some service information ( context depth, run temperature ).</li>
<li>https://api.openai.com/v1/threads/thread_id/runs/run_id (GET) – the call is used to get the status of the run ( to understand that the run is complete and we can see the results of the run ).  The “thread_id” parameter in the URL is the identifier of the thread that was created using the v1/threads call. The run_id parameter is the id of the run that was created using the /v1/thread_id/runs (POST) call.  Also, as part of this call we pass your API key.</li>
<li>https://api.openai.com/v1/models (GET) – the call is used to retrieve a list of AI models that are available for your account. This is a rare and optional call that is practically not used by the plugin. Also, within this call we pass your API key.</li>
</ul>
<p>More information about the OpenAI platform: <a href="https://platform.openai.com/" rel="nofollow ugc">homepage</a>, <a href="https://openai.com/policies/" rel="nofollow ugc">tems and policies</a>, <a href="https://openai.com/policies/terms-of-use/" rel="nofollow ugc">terms of use</a>, <a href="https://openai.com/policies/privacy-policy/" rel="nofollow ugc">privacy policy</a>.</p>


Talk To Website

AI chatbot powered by ChatGPT from OpenAI. Scans the content of your site, then trains an AI assistant that speaks to users like a live person.

WisdmLabs

<p>Your AI assistant is smart — but it knows nothing about your specific WordPress site. AI Conduit fixes that. Once installed, your AI can look up your blog posts, check your WooCommerce orders, pull student progress from LearnDash, or summarise your site’s data — just by asking, in plain language.</p>
<p>Think of it as giving your AI a secure window into your WordPress database. It can read and report. It can never edit, delete, or change anything.</p>
<h4>WordPress 7.0 has AI infrastructure — so why AI Conduit?</h4>
<p>WordPress 7.0 shipped the Abilities API, AI Client, Connectors screen, and MCP Adapter. For the first time, WordPress has a standardised way for plugins to talk to AI assistants. The architecture is sound. The vision is right. And for most WordPress sites today, none of it does anything useful yet.</p>
<p>Here’s why. The Abilities API is a registry — it only matters if plugins populate it. WordPress 7.0 ships with two baseline abilities. WooCommerce is leading the way, but the vast majority of the plugin ecosystem hasn’t registered anything yet. Until they do, your AI can see the registry, but there’s nothing in it for your site.</p>
<p>The bigger gap is your custom work. Most production sites have custom post types, meta fields, and bespoke integrations that no plugin author can register on your behalf. The only native path to exposing them is custom PHP — writing schemas and permission callbacks for every field you want your AI to read. That’s a developer project, not an afternoon’s work.</p>
<p>AI Conduit closes both gaps. Point and click to give your AI read access to anything in your database — standard fields, custom post types, custom meta, even data added by plugins that haven’t registered abilities yet. No PHP. No waiting on someone else’s roadmap. Working in minutes, not months.</p>
<h4>What can you use it for?</h4>
<ul>
<li><strong>Customer Support</strong> — Ask your AI “what did customer John order last week?” and get the answer without opening WooCommerce. Save hours every week.</li>
<li><strong>E-commerce</strong> — “Show me all orders over $5,000 this month.” Your AI assistant pulls the data, you make the decision. No exports, no spreadsheets.</li>
<li><strong>Content Teams</strong> — “List all posts published in the last 30 days with fewer than 500 words.” Your AI does the audit in seconds — not hours.</li>
<li><strong>Course Creators</strong> — With LearnDash, ask “which students haven’t completed Module 3?” and follow up personally — without running a single report.</li>
<li><strong>Site Owners</strong> — “Give me a weekly digest of new signups, published posts, and comments.” Your AI becomes a personal briefing assistant for your site.</li>
<li><strong>Agencies</strong> — Build tools once. Use them across client sites. Pull the metrics your clients care about without touching phpMyAdmin or writing SQL.</li>
</ul>
<h4>How it works — three simple steps</h4>
<ol>
<li><strong>Install and activate the plugin.</strong> Upload to WordPress the same way you install any other plugin. Takes about two minutes.</li>
<li><strong>Build your first tool with drag-and-drop.</strong> Choose what data your AI can see — posts, orders, users, comments. No code required. Point, click, configure.</li>
<li><strong>Connect your AI and start asking questions.</strong> Paste one connection URL into Claude Desktop or your preferred AI app. Your assistant now knows your site.</li>
</ol>
<h4>Your data stays safe</h4>
<ul>
<li><strong>Read-only access</strong> — Your AI can never edit, delete, or publish anything on your site. Ever.</li>
<li><strong>Role-based controls</strong> — Decide exactly which WordPress roles can access which tools.</li>
<li><strong>Full request log</strong> — Every query your AI makes is logged so you always know what was asked.</li>
</ul>
<h4>Everything included</h4>
<ul>
<li><strong>Visual tool builder</strong> — No code. Drag blocks to define what your AI can access.</li>
<li><strong>Rich data sources</strong> — Posts, pages, users, comments, and custom fields — all queryable.</li>
<li><strong>Smart filters</strong> — Narrow by date, category, author, custom field value, and more.</li>
<li><strong>Cross-data queries</strong> — Combine posts with their authors or comments with their posts in one query.</li>
<li><strong>Import & export tools</strong> — Save your tool configuration as JSON and reuse it across sites instantly.</li>
<li><strong>Rate limiting</strong> — Set caps on how often the AI can query your site to prevent overload.</li>
</ul>
<h4>Who is this plugin for?</h4>
<ul>
<li><strong>WooCommerce store owners</strong> — Query orders, customers, and product data without leaving your AI chat.</li>
<li><strong>LearnDash course creators</strong> — Track enrolments, completions, and quiz scores through natural conversation.</li>
<li><strong>Content editors & bloggers</strong> — Audit your content library, find gaps, and surface your best posts with ease.</li>
<li><strong>WordPress developers</strong> — Build custom MCP tools for clients. Extensible with your own block types.</li>
</ul>
<h4>Works with your favourite AI tools</h4>
<p>AI Conduit connects to any AI app that supports the Model Context Protocol (MCP). That includes:</p>
<ul>
<li>Claude Desktop</li>
<li>Cursor</li>
<li>Any MCP-compatible client</li>
</ul>
<p>New AI tools are adding MCP support every month — any compatible client will work automatically.</p>


AI Conduit

Give your AI assistant a secure, read-only window into your WordPress site. Works with Claude Desktop, Cursor, and any MCP-compatible AI.

Talk To Website vs AI Conduit