Premier Roofing engaged New Clarity to build an AI-powered application that allows business users to ask questions about company performance and receive immediate answers directly from their data warehouse.
The goal was to eliminate delays caused by BI processes, reduce dependence on technical resources, and enable faster, more flexible decision-making across the organization.
New Clarity developed a chat-based AI system that translates natural language into SQL queries, returns structured outputs, and allows users to explore data conversationally without requiring technical expertise.
Before this system, Premier operated in a traditional BI model where access to data was fragmented and slow.
Users did not have a single place to explore information. One-off analysis required exporting data into Excel and manually combining datasets. Shared reporting required submitting requests to the BI team, waiting for dashboards to be built, and receiving static outputs that answered only predefined questions.
This created several constraints:
Insight generation typically took about a week
Users could not explore data beyond predefined dashboards
Follow-up questions required restarting the process
Technical teams were constantly interrupted by ad hoc requests
A major inefficiency was context switching. Leaders would ask a question while actively working through a problem, but by the time they received an answer, the context was lost and the insight was less valuable.
At the same time, Premier’s technical team was forced to balance multiple roles. Data engineers were responsible not only for maintaining the data warehouse, but also for answering questions, validating reports, and troubleshooting discrepancies. In many cases, answering a single question required hours of investigation.
New Clarity built an AI-powered chat system that enables direct interaction with Premier’s data warehouse through natural language.
The system translates business questions into SQL queries using Premier’s schema, definitions, and example query patterns.
It was trained on approximately 50 real business questions paired with validated SQL queries and expected outputs, ensuring that responses align with how Premier defines key metrics such as revenue, margin, and performance.
This allows users to ask complex analytical questions without writing SQL or relying on dashboards.
The system maintains context across interactions, allowing users to ask follow-up questions without restating the full query.
Users can move from high-level questions to detailed breakdowns in a single workflow, enabling continuous analysis rather than isolated queries.
The platform returns results in multiple formats:
Direct numeric answers
Structured tables for multi-row outputs
Inline charts automatically generated based on the data
Exportable datasets for detailed analysis
All visualizations are generated directly from the underlying query results, ensuring consistency between reported data and visual outputs.
The system includes role-based permissions that control which data each user can access.
Users are assigned roles tied to specific datasets or tables, enabling broad adoption across the organization while maintaining appropriate data access controls.
The system improves through real-world usage and feedback.
Premier identified that the primary factor influencing performance is not the underlying technology, but the time invested in training and refining how the system interprets business questions.
As users interact with the system, it is iteratively improved to better handle real-world phrasing, edge cases, and business-specific logic.
The system reduced time-to-insight from approximately one week to immediate answers.
Business users can now ask questions and receive results in real time, without relying on dashboards or technical teams.
Key outcomes include:
Elimination of most manual SQL queries and spreadsheet-based analysis
Reduced dependency on the BI team for routine reporting
Faster, more flexible decision-making across the organization
Significant reduction in context switching during problem-solving
Increased access to data for non-technical users
The system is particularly valuable for mid-level operators and managers who have frequent analytical needs but limited access to BI tools.
Technical teams also benefit by regaining focus.
Instead of being interrupted by ad hoc data requests, they can concentrate on higher-value initiatives such as automation and system improvements.
This project fundamentally changes how Premier interacts with its data.
Instead of relying on static dashboards and centralized reporting, the organization can now access insights through a flexible, conversational interface.
The system establishes a clear path toward:
Replacing traditional BI tools with a unified interface
Consolidating dashboards and ad hoc analysis into one system
Reducing operational overhead in data reporting
Enabling faster and more consistent decision-making
A key next step is integrating standardized dashboards and conversational querying into a single experience, allowing users to view metrics and immediately ask follow-up questions in the same environment.
Over time, the system is expected to evolve toward proactive insights, where it not only answers questions but also surfaces patterns and recommendations automatically.