About the Client
The client is an advanced artificial intelligence solutions provider focused on designing and deploying intelligent systems, helping organizations operate efficiently and make smarter decisions. With strong capabilities across machine learning, data analytics, natural language processing, and automation, the company bridges deep technical research with real-world business applications. They are focused on building scalable and secure AI solutions that address complex operational challenges while delivering measurable impact. Driven by continuous innovation, the client emphasizes future-ready technologies that adapt to changing business needs and support long-term digital growth.
Business Challenges
The client faced challenges in accessing manufacturing data efficiently, as extracting insights required technical expertise to write complex SQL queries. The client manages a large portfolio of electrical and industrial products with highly technical specifications distributed across multiple data sources like product catalogs, datasheets, Excel files, and ERP systems.
Traditional keyword-based search and filter-driven systems are unable to interpret natural language queries, handle inconsistent terminology, or perform multi-attribute reasoning across documents. As a result, product identification requires significant manual effort, and operations teams depend heavily on IT or data specialists. Manual querying increased delays and the risk of errors, which slowed down operational insights. The client needed a solution that could understand simple, conversational questions and deliver accurate, real-time answers without technical intervention.
Kasmo’s Solution
We implemented an agentic solution built on Snowflake Intelligence to enable agentic AI driven insights across the client’s manufacturing data. The solution enables business users to ask questions in simple conversational language and receive accurate answers without writing SQL. We used Snowflake Cortex Analyst for structured data analysis and Cortex Search for unstructured and variant data, which ensures comprehensive insight retrieval across data types.
An AI Agent interprets each user query and intelligently determines whether to use Cortex Analyst or Cortex Search based on intent. Cortex Analyst automatically generates optimized SQL queries to fetch precise data from Snowflake, eliminating manual querying and reducing reliance on technical teams. Prompt-based customization defines agent behavior, tool selection, and response formats, ensuring consistent, business-friendly outputs aligned with the client’s operational needs. Enabled agentic rules and governance within Snowflake Intelligence, including tool selection logic, response behaviour, and standardized output formatting to ensure consistent and reliable AI driven insights.
Key Results Achieved
The Snowflake Intelligence-based solution significantly improved data access and analytics efficiency for the client. Key outcomes include:
- Integrated and transformed JSON and variant data into structured Snowflake tables for scalable querying and analysis.
- Organized and normalized manufacturing data to improve consistency and analytical accuracy.
- Implemented Snowflake Cortex Analyst by defining table relationships and semantic analysis.
- Customized Cortex Analyst using prompt engineering to align SQL generation and AI driven insights with business needs.
- Enabled Cortex Search for context-aware retrieval of insights from unstructured data.
- Built an AI Agent using Snowflake Intelligence that selects the right tool based on user intent.
- Delivered the Agent API and explained the full workflow, enabling easy client integration and testing.
