Preface
I have recently been thinking about how to use AI-related plugins on WordPress to achieve more features, but the trouble is, the vast majority of AI plugins are paid, making it hard for me to even try them out. On the other hand, if they are paid plugins, I estimate that most personal webmasters won't have much interest (including me), so even writing tutorials wouldn't make much sense.
Fortunately, after searching for a long time, I finally found a plugin that has a free version and offers plenty of basic features: AI Engine. It provides many features including site chatbots, text-to-image, image-to-text, article title generation, and article summaries. Crucially, I noticed there are basically no detailed Chinese tutorials for this plugin, so I chose it this time!
Introduction to AI Engine
AI Engine is a powerful WordPress plugin designed to enhance website functionality through AI technology, particularly by integrating OpenAI (such as GPT-3 and GPT-4) models. This plugin features a variety of practical functions, excelling especially in content creation and automation.
Main Features:
- Automatically Generate Content：
AI Engine can help users automatically generate blog posts, article summaries, product descriptions, SEO-optimized content, and more. You only need to enter keywords or topics, and the plugin will automatically generate relevant high-quality text (how high?) based on the input.
- Interact with ChatGPT：
The plugin provides integration with various large language models, allowing direct interaction with AI in the WordPress backend to quickly get inspiration and writing suggestions; or communicating with users on the website frontend, automatically replying to comments and messages based on user needs.
- AI Training：
AI Engine allows users to train their own AI models to generate more personalized content based on the specific needs of the website. This is very useful for websites that require a specific tone or style.
- Multi-language Support：
The plugin supports multiple languages and can automatically translate content, helping international websites create multi-lingual versions of pages.
- Fast Page Generation：
Through simple user input, AI Engine can quickly generate new pages and posts, improving the efficiency of website content updates (use this with caution; if the content is not good, just updating quickly doesn't make much sense).
Applicable Scenarios:
• Blog content creation
• Automatic generation of product descriptions
• AI auto-replies for customer interaction
• SEO content optimization
AI Engine's versatility makes it a powerful assistant for improving the productivity and user experience of WordPress websites, especially for sites that require frequent content creation.
AI Engine Plugin Installation and Settings
Install AI Engine
Install the AI Engine plugin:
Enable the plugin:
Introduction to AI Engine Settings
Settings Tab
“The ”Settings" tab is used to configure and manage various parameters of the system to optimize AI performance according to specific needs and scenarios. You can access the AI Engine settings interface from either of the two entry points shown in the image below:
Environments for AI
“The options in the ”Environments for AI" section are used to set the key information of the large language model providers required by each feature of the entire plugin. All model versions to be used must be configured here (including features like chat, text-to-image, image-to-text, voice, etc.):
AI Engine supports the following large model providers:
I chose Azure OpenAI because it can be accessed in China without needing ”science" or "magic" (VPNs). If you choose other large model providers and your WordPress host happens to be in China, then individual webmasters will need to use "science" or "magic" to ensure that WordPress can normally access the API addresses provided by those other large model providers.
The ”Deployment Name” and ”Model Version” involved in the ”Deployments” section need to correspond to the configurations in the large language model provider. Taking this article as an example, the following ”Deployments” is my final configuration:
For the content to fill in ”API Key” and ”Endpoint” in the image above, please refer to the ”Keys and Endpoint” section of Azure OpenAI:
Meanwhile, the ”Deployment name” and model version must match the ”Deployment name” and ”Model name” in the ”Deployments” section of Azure OpenAI:
The reason they must match is that when AI Engine makes requests to the Azure OpenAI API, it uses this information to construct the request URL.
Default Environments for AI
“The options in the ”Default Environments for AI" section are used to configure the default environment and behavior of the AI at runtime, setting default AI environment parameters to ensure the AI performs as expected in different use cases, as shown in the image below:
As its name suggests, GPT-4o-mini is a smaller version of the GPT-4 model, optimized for specific tasks (such as conversational AI). Typically, models like GPT-4o-mini are designed to prioritize chat-based interactions rather than advanced multimedia features like text-to-image (DALL-E) or image-to-text (CLIP) generation.
These advanced features, such as generating images from text or processing images into descriptions, are typically handled by other specialized models (for example, DALL-E for image generation, CLIP for image understanding, Whisper for speech-to-text, tts-1 for speech synthesis, etc.). Therefore, GPT-4o-mini primarily focuses on language understanding and generation, making it suitable for tasks like chatting, writing, and answering questions, but not for multimedia generation.
Consequently, Images, Embeddings, and Audio cannot use GPT-4o-mini (which is GPT-4). If you want to set the default AI for these features, you must first add dedicated models like DALL-E-3(h) or CLIP in the preceding ”Deployments” section.
General and Chatbot
“The ”General” section is used to configure global settings of the system, such as language, performance, and error handling, to ensure stable AI operation; the ”Chatbot” section focuses on the interactive behavior of the chatbot, covering conversation style, welcome messages, context management, etc., to enhance the personalization of the user experience and conversational coherence. The functions of the options in the General section and Chatbot section are shown in the image below:
Admin Tools
Admin Tools is used to specify whether to pin these 4 feature entry points to the top bar of the admin dashboard:
Discussions Tab
“The ”Discussions" tab is typically used to manage and view interaction records with users in the system. Its functions include the following aspects:
- Conversation Management: This tab helps you view historical conversations with users, making it easy to quickly review and reference. For complex issues or multiple conversation topics, the “Discussions” tab allows you to systematically track each user's needs and questions.
- Contextual Association: By viewing previous discussions, the AI can provide more accurate replies in subsequent conversations, ensuring conversational coherence and accuracy. Users” needs may become clearer over multiple interactions, so the ”Discussions" tab helps to better understand the overall context.
- Quality Improvement: Using the ”Discussions” tab can help train the AI model, improving its response capabilities by reviewing and analyzing previous conversations. This also helps identify common user needs or questions, thereby further optimizing the system.
Meanwhile, the ”Discussions” tab is also the content that appears after the Discussions option is set to ”Enable” in the “Settings” - ”Chatbot” section:
Chatbots Tab
“The options under the ”Chatbots" tab allow users to configure the chatbot's model, initial prompt, AI replies, context management, user input handling, error handling, custom commands, file uploads, and appearance settings. These features allow users to flexibly customize the chatbot's behavior and interface.
Mode
AI Model
This option depends on the selection in the previous ”Mode” section. If Chat is selected, then choose ”GPT-4o Mini” here; if Images is selected, then ”DALL-E-3(H)” should be selected here:
Thresholds
This section can be used to limit the content a user can input at one time:
Appearance
Set the appearance, content, avatar, etc., of the chatbot interface:
When the avatar is unchecked:
When the avatar is checked:
**Item 9:**Gravatar is a globally used avatar service, standing for Globally Recognized Avatar. It allows users to associate an avatar with their email address. When users use the same email address on websites that support Gravatar (such as WordPress, forums, etc.), Gravatar automatically displays the avatar bound to that address. Users only need to upload their avatar once on the Gravatar platform, and it will automatically display on all Gravatar-supported websites thereafter. This service simplifies the process for users to set up personal avatars across multiple websites, providing a unified online identity. The Gravatar website is as follows:https://gravatar.com/, those who need it can go and register.
Other Options
- Functions: You can extend the features of the AI model by installing ”Snippet Vault”. It will use these features to enhance responses and perform specific actions; this functionality is known as function calling or tools.
- Shortcodes: You can implement chatbot features in specific locations using shortcodes.
- Actions: You can copy or reset the configuration of the chatbot section.
- Themes: The appearance of the chat section can also be manually changed:
- Site-Wide Chatbot: By default, the chatbot is only available in the WordPress admin dashboard. You can change the value of the Site-Wide Chatbot option (top right corner) from ”none” to ”default”, and the chatbot avatar will then appear in the bottom right corner:
Click to open and display the chat dialog box:
However, enabling it globally means anyone can use your chatbot, which costs money, so please be careful~ Generally, it is used by businesses to communicate with prospective clients; for individual webmasters, feel free to use it if you are wealthy.
Dashboard
Client Modules and Server Modules. For the Server Modules section, select based on your actual needs; do not check them all for no reason, as it will increase WordPress resource consumption:
Admin Modules:
Usage:
Using AI Tools
As mentioned earlier in the article, in AI Engine, you can display the three AI tools ”Content”, ”Images”, and ”Playground” in two locations on the WordPress dashboard through the ”Admin Tools” and ”Admin Modules” options:
Content
“The ”Content" feature helps users generate, edit, and optimize text content. You can use it to create articles, blogs, marketing copy, or any type of text output. It can generate targeted and high-quality content based on the topics, keywords, or style requirements provided by the user. Additionally, it can improve existing text, such as polishing, correcting, simplifying, or expanding it. This feature is suitable for rapid brainstorming, content optimization, and improving writing efficiency. See the image below for an introduction to the tool's interface:
The effect is as follows:
Of course, this kind of fully automatically generated content is basically unreadable, so we can further refine the content by manually filling in titles and section names:
The more detailed the prompts, and by manually specifying accurate titles, section names, and excerpts while only automatically generating the body content, the more the content will meet your expectations.
Images
“The ”Images" feature can generate custom images based on user descriptions. You can input a detailed text description (called aPrompt), and the AI will generate corresponding images based on this description. Users can specify image sizes, such as square or rectangular formats, and can also choose to generate one or multiple images. The “Images” feature is suitable for creating visual content, such as designing concept maps, artworks, or displaying specific scenes. This feature allows flexible adjustment of details to ensure the output meets user expectations.
To use the image generation feature, you need to first add a DALL-E series model that supports image generation in ”Settings”. Here, I will take 3 as an example:
After that, you can generate images in the Images tool. The introduction to the tool's operation interface is shown in the figure below:
Playground
“The ”Playground" tool is a feature-rich interactive platform that allows users to select and test various AI models, adjust parameters, conduct prompt engineering experiments, and view results in real time. It supports multi-turn dialogue, code generation, multi-language processing, and provides API integration examples. This tool is not only suitable for developers and researchers to conduct technical exploration and optimization, but also provides content creators and educators with an intuitive environment for learning and demonstrating AI technology. See the image below for how to use the tool:
Note: In fact, the so-called templates are just some detailed prompts, so you can also refer to the 4 default templates, write down some prompts you frequently use in advance, and save them as templates for easy reuse in the future. Below is a demonstration of actual use:
The result is not bad:
Other Integration Features
The integration features here mainly refer to the features integrated into the backend article section interface, including naming articles and generating article summaries. Hovering the mouse over any article will reveal two features starting with a ”magic wand” at the end of the menu items: ”Title” is for naming the article, and ”Excerpt” is for generating an article summary:
Click ”Title” to generate recommended article names:
Click ”Excerpt” to generate multiple article summaries for you to choose from:
Moreover, in addition to these 2 features, the Images tool will also be integrated soon:
Note: The principle of generating article summaries is to send the article content to the AI, and then the AI will generate a summary based on the received content. This will consume a large number of sent tokens. If it is a long article, it feels like a loss, so please manage this feature yourself. Well, OpenAI 3.5 is free. If you can use OpenAI, you can try this method (build an API reverse proxy yourself using a worker), I won't bother. Although Azure OpenAI can be accessed without VPN/proxy and has no monthly fee (pay-as-you-go), it has no free quota either. Indeed, you can't have your cake and eat it too.
Summary
I think the biggest advantage of AI Engine is that the free version provides enough features and is quite practical. Compared to other plugins that easily cost $7 or $8 a month, it is already very generous.
However, except for the chatbot, other features of AI Engine are actually just optional. ”Whether they are useful and how to use them” is a matter of opinion: for example, if I want to generate text, images, or article summaries, why should I generate them in the WordPress backend? Can't I just use ChatGPT or other models to generate them locally and then put them into WordPress? Generating them on WordPress also consumes the performance of the host server~.
Therefore, everyone can choose what suits them based on their actual needs.
P.S.: Although I have tried to introduce these options in as much detail as possible, there are still errors and omissions. It seems there are a few I didn't mention. Forget it, I'm too lazy to add them, and they don't seem to be important options anyway.