When do I need AI Search Attributes?
Use them for recurring exact facts that customers may ask about directly, such as instructor, brand, branch, level, or duration. Instead of relying only on text similarity, you give Mr. AI structured information it can use for precise lookup, filtering, and counting.
- Attribute = the type of fact, such as Instructor.
- Value = the fact attached to an item, such as Hassan Faleeh.
- Aliases and Search Name help the system recognize that same value across different wording.
Supported value types
- Text: an exact text fact such as an instructor or branch.
- Enum: a value from a known recurring set such as Level or Category.
- Number: a numeric fact such as Duration or Capacity when supported comparisons matter.
- Boolean: a clear yes/no fact such as Online.
What is the difference between Value, Alias, and Search Name?
- Value is the main value that represents the fact inside Catalog.
- Alias is another way to write the same value, not a different value. For example, حسن فليح and Hassan Faleeh can be aliases when both identify the same person.
- Search Name is a stable machine-friendly name that helps the system connect the value to different language or spelling forms.
- Search Name does not replace the customer-facing name; it helps retrieval understand what the name refers to.
Search Name format
- Use lowercase English letters.
- Use a single space between words.
- Do not use digits, underscores, hyphens, punctuation, or uppercase letters.
- Example: أمنية علي عامر can use omnia ali amer as its Search Name.
- Example: العربية can use arabic when that is the intended value.
How does Mr. AI use it?
The system first tries to resolve what the customer actually wrote. If that value is not clear enough, Search Name can help connect it to the same entity through a stable form. If the result is still ambiguous, the system should avoid forcing a precise filter that could return the wrong item and can fall back to the safer semantic path.
Attribute or Semantic Tag?
- Use an Attribute for an exact fact that can support direct lookup, filtering, or counting.
- Use a Semantic Tag for flexible discovery meaning that is not maintained as an exact business field.
- Do not turn AI inference or assumptions into hard Attributes without a reliable source.
What about older items?
Items without AI Search Attributes continue to be searchable and usable. Add Attributes gradually where you need higher precision instead of trying to convert the entire Catalog at once.
Updating or disabling an Attribute
Definitions can be changed or disabled. When an Attribute already has values attached to items, use the safer Disable path when the product requires it instead of removing existing relationships unexpectedly.