Performing Analyses
For everyday users
Last updated
For everyday users
Last updated
With your Database and Knowledge Base integrated, you are ready to request your first analysis. Depending on the type of request, Patterns AI has two separate modes for generating outputs.
Patterns will perform an autonomous multi-step analysis, generating potentially multiple SQL queries and charts to answer a users request.
However, if a specific Report is requested, Patterns will follow the specified template to generate the report and analyze the resulting data.
This guide provides detailed documentation of the system's features, with direct links to relevant moments in the video demonstration.
The interface allows users to ask natural language questions, which the AI interprets to generate SQL queries and insights. For example, the user asks, “What are the best-selling products over time?” The AI generates steps to retrieve and visualize this data, including grouping sales by product ID and aggregating over a time period. Watch this feature in action at 00:00:00.
The AI generates visualizations to represent query results, such as time-series charts for sales trends. Charts can be customized by limiting results (e.g., top 10 products) for clarity. This process is shown at 00:01:06. Users can further customize these charts by editing Vega-Lite specifications directly, seen at 00:04:50.
The "Edit Message" feature lets users modify queries without restarting. For example, a query analyzing monthly sales was adjusted to analyze quarterly data, demonstrated at 00:01:52. The interface supports running multiple iterations simultaneously, useful for side-by-side comparisons.
Analyses can be saved to a global repository for reuse. Saved analyses include the SQL query, data, and visualization. Learn how to save an analysis at 00:03:06. Saved items can also be added to a shared knowledge base for collaborative purposes.
Users can modify the SQL queries generated by the AI or edit the Vega-Lite specifications for visualizations. For example, a line chart was modified into a bar chart, as shown at 00:04:50. These features provide precise control over the analysis and visualization outputs.
The interface includes tools for collaboration, such as tagging teammates and leaving comments. In the demonstration, a user tagged a colleague to review an analysis at 00:06:38. Notifications are sent via email, and permissions can be managed to control access levels.
Users can upload external files like CSVs to integrate with the existing data warehouse. At 00:05:52, a trial balance file was uploaded and analyzed alongside warehouse data. This feature is useful for tasks such as reconciliation or supplementing existing datasets.
Visualizations and data can be exported for use in presentations or reports. Charts can be downloaded as image files, while data can be exported as CSV or Excel files. This process is shown at 00:04:27.
Chart Update Issues: Changes to chart types may not apply as expected. This issue, demonstrated at 00:05:12, can be temporarily resolved by editing the Vega-Lite code directly.
Unexpected Query Results: If queries do not match the intent, refine the question or adjust the SQL manually.
Start with broad queries and refine iteratively.
Save analyses for reuse and collaborative purposes.
Use customization options for precise control over outputs.
Collaborate effectively by tagging team members and managing permissions.
Regularly export visualizations and data for offline use.
