When you install MagicAI, you’ll arrive at the app’s Home page. Here, you’ll find a quick onboarding guide, a video tutorial on how to use the app, your subscription details, and the available credits for content generation.
The Home page also features an Analytics dashboard with two charts:
- Daily number of generations over the last month
- Total generations over the past 90 days.
Additionally, you’ll see the Changelog and quick-access buttons to rate the app, purchase credits, or view your plan.
Before you start generating content, we recommend specifying Settings for the application to ensure everything is configured correctly.
General Settings
Open Settings page and specify the following:
Preview before publishing - enable the preview functionality to review the generated content during mass action before publishing.
Append images to generated content - set Yes to add current content images (if any) to the end of the generated description. The option also works for description refinement - in this case, images remain in their original positions.
Generation strategy - select the generation strategy that is used when an error occurs with the overloaded OpenAI service.
Model - three distinct language models with its own set of capabilities, select the one that best suits your specific language requirements.
Note
The GPT-4o (available from the Basic plan) - high-intelligence model for complex tasks.
The GPT-4o Mini - intelligent small model for simple tasks.
Format - this parameter allows you to select the desired format for the generated content.
- By selecting the "HTML" option, the content generator will create text that adheres to the HTML markup language. This enables the creation of structured content, including headings, paragraphs, lists, links, images, and other web page elements.
- By selecting the "Plain text" option, the content generator will produce unformatted text without any styling or markup. This simple text format does not include styles, fonts, or external elements.
Language - specify the language used for text generation.
Length - choose the size of the generated text. It determines the number of tokens in the output generated by the model:
- Short - choosing the short length option limits the generated text to a maximum of 300 tokens.
- Standard - the standard length option allows the generated text to be up to a maximum of 600 tokens.
- Long - selecting the long length option allows for more extensive and detailed responses, with a maximum of 800 tokens.
Tone - tone refers to the desired emotional or stylistic tone of the generated text.
Exclude Title - remove the title from the generated content. This ensures that the product name doesn't duplicate in both the title and the content.
Temperature - specify the level of randomness and creativity in the generated text.
- Higher Temperature: Setting a higher temperature, such as 0.8 or 1.0, makes the output more random and creative.
- Lower Temperature: Setting a lower temperature, such as 0.2 or 0.5, makes the output more focused and deterministic
Note
These settings are the basis for the content generation, both for individual content and multiple content through the 'Generate description'/'Generate SEO' mass action. However, it is important to note that these settings can be individually modified for specific object if custom adjustments are required.
After configuring the settings, switch to the necessary object grid, where you can generate content individually or in bulk.
Note
The app uses credits for content generation: 1 credit = 1 description or SEO item. For bulk SEO generation, 2 credits are needed for each SEO item (1 for the SEO title and 1 for the SEO description).
Products content generation
Open Products page.
The grid provides sorting, searching, and filtering by status functionality for a better user experience.
Generate content for products in bulk by selecting items and clicking the desired action:
- Generate Description - create descriptions for selected products.
- Generate SEO - produce meta titles and descriptions (available from Basic plan).
- Refine - improve the grammar, syntax, and wording of your existing product descriptions.
- … > Generate descriptions with custom prompt - use this option to provide your specific instructions to ChatGPT for product description generation - either add your own prompt manually or select custom prompt template from the dropdown.
When using a custom prompt, make sure to use the global.input variable. You can find and copy it by going to Prompts > Add new prompt in the app’s admin panel. This step ensures that ChatGPT accesses information about your product from your product settings. If you're looking for ideas for custom prompts, check out our blog post.
Note
Here’s how credit usage is calculated for generation:
1 product description = 1 credit
1 SEO = 2 credits (1 for SEO title and 1 for SEO description)
1 refine = 2 credits
Stop generation - click the button to instantly cease mass generation whenever required.
You can also generate content for each product individually for a more tailored experience.
For it, open a product you want to generate content for.
You can generate product titles, descriptions, SEO titles, and SEO descriptions using the Generate or Generate with custom prompt buttons. The latter allows you to create custom descriptions with tailored prompts.
When using a custom prompt, make sure to use the global.input variable. You can find and copy it by going to Prompts > Add new prompt in the app’s admin panel. This step ensures that ChatGPT accesses information about your product from your product settings. If you're looking for ideas for custom prompts, check out our blog post.
If you already have a description or SEO data, simply use the Refine current button to improve the grammar, syntax, and wording of your existing description, SEO title, and SEO description.
If you want ChatGPT to consider the current product description when generating a new one (applicable only to the ‘Generate’ option in the Generate Description section), enable label. If you prefer ChatGPT to generate it from scratch, leave it disabled.
Note
This feature works differently from the "Refine" option. When generating a description with a label on, ChatGPT uses all the product-specific settings you’ve provided — such as keywords, title, category — and the information from the current description to create a new description. In contrast, the "Refine" option simply improves the grammar, syntax, and wording of your existing description.
Note
Here’s how credit usage is calculated for generation:1 product title = 1 credit
1 product description = 1 credit
1 SEO title = 1 credit
1 SEO description = 1 credit
1 refine = 2 credits
In the right column, you can customize specific product details such as Title, Type, and Category. These details help the AI generate accurate descriptions. You can also add Additional information and specify Keywords, either from a suggested list or your own. If available, you can Include product variants. Additionally, you can set the Length, Temperature, Tone, Language, and Format of the generated text.
Note
Product-specific settings (Products page) take priority over general settings (Settings page), allowing you to adjust parameters for each item as needed.
Save changes to publish.
Note
All published content remains saved even after uninstalling the app.
Collections content generation
Open the Collections page and use the search bar to find the collection you want to generate content for. Search results are based on matches in the collection’s title and description.
To generate content in bulk, select collections and click the desired action: Generate Description, Generate SEO, or Refine. To use your custom instructions to ChatGPT, click … > Generate with Custom Prompt and either add your own prompt manually or select custom prompt template from the dropdown.
On the collection customization page, you can enter additional information and customize settings for content generation. If you don't specify these settings, the general settings from the Settings section will be used.
With the application, you can generate and refine the following data for collections:
- Content;
- SEO title;
- SEO description.
Note
Here’s how credit usage is calculated for generation:
1 collection description = 1 credit
1 SEO title = 1 credit
1 SEO description = 1 credit
If you want ChatGPT to consider the current collection description when generating a new one, enable the label (applicable only to the ‘Generate’ option in the Generate Content section). If you prefer ChatGPT to generate it from scratch, leave it disabled.
Click Generate with custom prompt to provide specific instructions to ChatGPT for tailored content. In the dialog window, enter custom instructions or select a ready-made prompt template from the dropdown.
Blog posts content generation
Go to the Blogs page and select the blog where you’d like to create a post.
Note
A blog page is needed to publish your articles. If you don’t have any Blog pages yet, you can create it by clicking the Create Blog button. You’ll be redirected to Shopify’s Blog settings, where you can add the blog title, SEO details, manage comments, and select a theme template. Save your changes, then return to the app.
When the blog page is selected, click Create Post. This will redirect you to Shopify’s Blog settings, where you can add the post title, an excerpt (post summary), SEO details, visibility settings, a featured image, and other options. Save changes.
Return to the app to complete your post setup (Blogs > open blog page > open newly created blog post).
Here, you can Generate blog post content automatically or, to add your own instructions, click Generate with custom prompt and enter your prompt. You can also Refine current content or Generate a featured image for the blog post.
If you're not satisfied with the result, simply click Regenerate or Regenerate with custom prompt to try again. Save changes to publish them.
Note
Here’s how credit usage is calculated for generation:
1 post generation = 3 credits
1 refine = 6 credits
1 image generation = 20 credits.
Custom Prompt Templates
If you prefer using custom prompts for content generation, you can save frequently used prompts as templates to easily reuse them.
Open Prompts. If you don’t see Prompts in the navigation menu, click View more.
To create a new template, just click Add new prompt button and fill-in all necessary details. 
Make sure to use the global.input variable to ensure that ChatGPT accesses information about your product from your product settings. If you're looking for ideas for custom prompts, check out our blog post.
Note
Default prompt template is not editable. Please use the Duplicate function to create a new template and make the necessary changes.
When generating content for an item, select Generate with custom prompt and choose the appropriate template from the dropdown.
History
If you want to restore the original content or revert to a previous version, simply go to the History page.
On the History page, you can view a timeline of your generation history. If you don’t see History in the navigation menu, click View more.
The search is conducted based on the generated content rather than the object for which the generation was performed. The history grid is sorted into Generation history and Published history. It allows you to effortlessly revert all publications back to their original versions. You can also filter by date for quicker access.
View content - click the link to view original and generated content.
Restore - click the link to recover previous versions of any content if needed.
Buy extra credits for content generation
If you need more credits to create content beyond your current plan, you can easily purchase extra credits in the app.
To do this, go to Pricing, click the link to top up 100 credits for just $1 in your pricing plan tab. Choose from 100, 250, or 500 credits based on your needs, and click the Buy button.
