Search

When you ask DataQI to find information, it does much more than simply look for exact keyword matches. The platform is equipped with an array of search tools that allow agents to use advanced methods to find exactly what you need.

Crucially, security is front and centre: all search operations are strictly bound by user-scoped connection authentication. The search tools run using your active permissions, ensuring you only ever see search results for files or directories that you naturally have permission to view in the underlying source system.

How DataQI understands search

DataQI helps you find the right thing, quickly. It does this by performing multiple types of search across all ingestion tiers simultaneously. While certain queries might lean heavily on specific tools, the platform seamlessly blends these strategies:

• Known-item search: For specific file lookups (e.g., “Find the Q3 Financial Report PDF”), DataQI runs targeted searches for maximum precision and speed.
• Semantic search: When exploring a concept (e.g., “Show me documents about improving network redundancy”), DataQI uses vector embeddings to find conceptually relevant information across document contents, even if your exact keywords don’t appear.
• Combining filters: For highly specific requests (e.g., “Find all sales contracts from 2024, excluding vacuum cleaners”), DataQI chains tools together. It generates a broad dataset and applies structured filters to narrow down the results.

Over time, DataQI employs smart indexing methods. The more interactions users have with the platform, the more accurate, inclusive, and detailed the search index becomes—always while strictly respecting file system access controls.

flowchart TD
    User([User Request]) --> Agent(Search Agent)
    Agent -->|Known-item| TS1[Metadata Search]
    Agent -->|Semantic| TS2[Vector Search]
    TS1 --> Results{Raw Results}
    TS2 --> Results
    Results -->|Filter Tool| Refined[Refined Dataset]

Interactive search results

When you ask DataQI to find files, it does not just return a wall of text. It generates an interactive search results table directly within the chat interface.

One of the key advantages of this approach is that by putting the search results into a structured table, the platform prevents the AI from getting confused or making up answers that may occur when large language models try to deal with massive search result sets.

This table is designed for both quick viewing and deep data manipulation:

• Dynamic columns: The agent intelligently selects which columns to display by default based on the context of your request.
• In-UI manipulation: You can sort and filter the search results locally within the table without needing to write a new prompt or wait for the AI to process a new request.
• File navigation: Every result includes a direct, clickable link to the source file (assuming you still have the correct permissions to access it in the source system).

Exporting large datasets

DataQI search is built to handle enterprise-scale repositories. While the chat interface will neatly paginate large results to keep your browser responsive, the table includes a highly visible Export to CSV function.

This allows you to download the full list of results (up to hundreds of rows) directly into external tools like Excel for offline manipulation.