AI assisted search
The AI assisted search provides you with a new way to interact with our documentation. Instead of browsing through multiple pages, you can now quickly find specific information or ask natural language questions to get immediate answers.
This guide explains how to make the most of the AI assisted search, its current limitations, and best practices to help you find the resources you need.
Searching for Specific Information
As you enter text into the search bar, the AI Search attempts to find matching content across documentation pages. To get the best results, try to match the page title or the content of the page as closely as possible.
Search results are limited by the specific version of documentation you’re currently viewing.
For example:
Searching for “Tokens” while viewing Package Security Manager Cloud documentation will return this Tokens page, but searching for “Tokens” while viewing an Package Security Manager (On-prem) version of documentation returns this Tokens page. Give it a try!
However, it’s important to understand that even exact terminology doesn’t guarantee perfect results!
For example:
If you are viewing the Package Security Manager (On-prem) documentation, a search for “conda-repo-cli” returns no results, but searching for “conda-repo-cli package” will show the Package Security Manager CLI documentation page. Alternately, you could search for “Package Security Manager CLI” or even just “CLI” to get the same search results.
This happens because you need to enter enough context to help the model determine what you are looking for. If you don’t get the results you expect, try rephrasing your search using terms from our documentation.
Asking questions
Pressing Enter/return submits your query to the AI model, which processes it as natural language and produces a response based on the model’s knowledge of the documentation. For example, if you want to know how to install Miniconda on a Windows machine, you can ask “How do I install Miniconda on Windows 11?”. The model will return a step-by-step process as an answer to your question along with a list of relevant pages for you to explore.
You can also ask follow up questions to refine your results. For example, if you ask “How do I install Miniconda” the AI search returns instructions for multiple platforms. You can continue your conversation by entering “On Windows” to get a revised response with instructions specific to Windows.
Current Limitations
Understanding the AI search’s limitations will help you use it more effectively:
Version-Specific Results
Search results generally only consider the documentation version you’re currently viewing. Always verify you’re in the correct product and version documentation before searching.
Terminology Sensitivity
The AI responds best to terminology that exactly matches our documentation. If you don’t get the expected results, try rephrasing using terms from our product pages.
Product Overlap
Because our documentation covers multiple products with overlapping features, you may occasionally receive information about a similar feature in a different product. Always check that the results match your specific product context.
Context Understanding
While the AI understands natural language, very complex or ambiguous queries may produce unexpected results. Break complex questions into simpler, more direct queries for better results.
Best practices
- When searching for specific information, try to match the page title or the content of the page as closely as possible.
- When asking questions, try to be as specific as possible. For example, include the product name and version to your search, if applicable.
- Make sure that you are viewing the documentation for the correct product and the correct version when asking questions.