Waii 1.30 introduces powerful automatic learning capabilities from enterprise data sources, significantly reducing the time needed to implement accurate conversational analytics in your environment.
Key Feature: Enterprise Data Source Integration
Our new release can now automatically learn from various enterprise sources:
- Documentation formats: PDF, text, HTML, JSON
- BI platforms: Looker ML, Power BI, Tableau workbooks
- Data catalogs: Incorporates metadata seamlessly
How It Works
Waii intelligently processes new information against your existing knowledge graph:
- New information: Creates semantic nodes that capture information concisely and links to all relevant objects
- Existing information: Skips duplicates to maintain a clean knowledge graph
- Enhancements: Captures only the delta of new information
- Conflicting information: Adds disabled nodes with explanatory notes, prioritizing existing information until manual review
Implementation Details
All captured information includes:
- Source tagging
- Unique IDs for each addition
- Simple interfaces for finding, editing, removing or reloading information
Why This Matters
Enterprise text-to-SQL accuracy depends heavily on contextual understanding, but manually generating this context is time-consuming. Most organizations already maintain documentation and semantic information in various systems.
Waii 1.30 lets you bring all this information together within minutes, dramatically accelerating your setup process and improving accuracy from day one.
Example: Learning Business Metrics
Original Power BI Definition
In Power BI, a company has defined "Adjusted Revenue" as follows:
Adjusted Revenue = SUM(Sales.Revenue) - SUM(Sales.Returns) - SUM(Sales.Discounts) *
IF(MONTH(Date) = 12, 0.85, 1.0)
Semantic Statement in Knowledge Graph
After processing the Power BI definition, Waii creates this semantic statement:
"Adjusted Revenue is calculated as the total revenue minus returns and discounts, with a special adjustment for December. Total revenue comes from the Sales.Revenue field. Returns are subtracted using the Sales.Returns field. Discounts from Sales.Discounts are also subtracted, but in December (Month=12), only 85% of discounts are applied to account for seasonal promotion adjustments. The metric requires Sales and Date tables to be joined. This definition was imported from PowerBI.RevenueAnalysis.pbix, measure ID DAX_Measure_291."
User Query Example
A business analyst later asks Waii:
"Show me adjusted revenue by month for 2024"
Generated SQL
Waii leverages its knowledge of the semantic statement to generate:
SELECT
DATE_FORMAT(d.date, '%Y-%m') AS month,
SUM(s.revenue) - SUM(s.returns) -
CASE
WHEN MONTH(d.date) = 12 THEN SUM(s.discounts) * 0.85
ELSE SUM(s.discounts)
END AS adjusted_revenue
FROM
sales s
JOIN
date d ON s.date_id = d.id
WHERE
YEAR(d.date) = 2024
GROUP BY
DATE_FORMAT(d.date, '%Y-%m')
ORDER BY
month
By seamlessly incorporating your existing business knowledge, Waii 1.30 ensures that your text-to-SQL functions leverage all the context and business logic you've already defined elsewhere in your organization.