Overview
Locate information within your stored documents using two powerful search methods: a dedicated search interface for structured queries, and conversational search through the chat interface. Both approaches access the same underlying data but serve different workflow needs—direct extraction versus AI-assisted interpretation.
Learning Objectives:
Access dedicated search from collections and datastore pages
Configure search parameters including result count and nested datastore inclusion
Execute conversational searches through the chat interface
Understand when to use each method based on your use case
Interpret search results with metadata and source information
Prerequisites
You'll Need:
Access to a workspace on Starter, Best-of-Breed, Professional, or Enterprise plan
AI Datastore collections with uploaded documents, webpages, or websites
At least one datastore containing searchable content
For dedicated search: Understanding of specific information you're seeking
For chat search: An active chat session
Note: The dedicated search feature is not available on Basic plans. Users on Basic plans can only search via the chat interface.
Method 1: Dedicated Search Feature
Access the Search Interface
From Collections Page:
Navigate to the left panel and click the "Collections" option
Locate your target collection in the collections list
Click the search icon (🔍) next to the collection name
From Datastore Page:
Navigate to the left panel and click the "Datastores" option
Locate your target collection in the datastore list
Expand the collection by clicking the arrow (▶️) if needed
Click the search field to find a data store
Result: The search page opens with pre-selected collection and datastore values based on your selection.
Tip: You can change the selected collection or datastore within the search page using the dropdown menus at the top of the interface.
Method 2: Conversational Search Through Chat Interface
Access Chat Interface
Navigate to the left panel and select any active chat or create a new chat
Navigate to the right sidebar by clicking the expand arrow (◀️) if collapsed
Scroll to the "Datastores" section in the right sidebar
Result: The right sidebar displays available datastore organized by collection.
Tip: You may need to click the "Datastore" tab or scroll down if using a mobile device or narrow window.
Select Datastore for Search
Click the checkbox next to the datastore(s) you want to search
Select multiple datastore if needed (AI will search across all selected)
Result: Selected datastore show a checkmark and are activated for the chat session.
Note: You can enable multiple datastore from different collections simultaneously. The AI will search across all selected content.
Execute Conversational Search
Click the message input field at the bottom of the chat
Type your natural language question referencing the content you need
Click on Internal to select your datastore
Press Enter or click the submit arrow to send the message
Example conversational queries:
"What are our Q4 sales figures from the revenue reports?"
"Summarize the key points from the product roadmap document"
"Based on our customer feedback data, what are the top complaints?"
Result: The AI searches selected datastore, analyzes content, and provides a conversational response with synthesized information.
Result: Responses include citations to source documents and may offer follow-up suggestions.
Warning: Chat search consumes more tokens than dedicated search because the AI generates interpretive responses. Monitor your token usage if on a token-limited plan.
Best Practices
Use dedicated search for direct quotes: When you need exact text from documents, use dedicated search to avoid AI paraphrasing.
Use chat search for synthesis: When you need understanding, summary, or analysis across multiple sources, use conversational search.
Enable multiple related data hubs: For comprehensive answers, select all relevant data hubs. The AI performs better with broader context.
Phrase questions clearly: In chat search, ask specific questions. "What was our revenue in March?" is better than "Tell me about revenue."
Configure appropriate result count: For dedicated search, set 10-20 results for most queries. Use 50-100 only when reviewing extensive related content.
Common Questions
Q: How many data hubs can I search at once?
A: Dedicated search queries one data hub at a time. Chat search can query multiple simultaneously—there's no hard limit, but performance is best with 3-5 selected.
Q: Can I search websites stored in data hubs?
A: Yes. Both dedicated search and chat search can query content from websites you've stored as data hub documents.
Q: Why do I get different results between the two methods?
A: Dedicated search shows raw matches based on keyword relevance. Chat search uses AI understanding and may synthesize information differently based on your question's intent.