Direct text-to-SQL is hard to trust in production because raw schemas do not capture governed metrics or business meaning. This post walks through a local, open-source proof of concept that puts a semantic layer between the language model and the lakehouse, combining Strands, WrenAI, Trino, Iceberg, and long-term agent memory.

The world of data is converging. The traditional divide between batch processing for historical analytics and stream processing for real-time insights is becoming increasingly blurry. Businesses demand architectures that handle both seamlessly. Enter the “Streamhouse” - an evolution of the Lakehouse concept, designed with streaming as a first-class citizen.

Today, we’ll introduce three key open-source technologies shaping this space: Apache Paimon™, Fluss, and Apache Iceberg. While each has unique strengths, their true power lies in how they can be integrated to build robust, flexible, and performant data platforms.

We'll discuss how to implement data warehousing ETL using Iceberg for data storage/management and Spark for data processing. A Pyspark ETL app will be used for demonstration in an EMR local environment. Finally the ETL results will be queried by Athena for verification.