There is no shortage of platforms for building applications infused with artificial intelligence (AI) but Snowflake is making a case for using one that already houses the data needed to train AI models. The company today announced at its Snowday 2023 conference that it is extending its support for Python workloads within its Snowpark framework for building and deploying applications.

At the same time, Snowflake is previewing a managed service to make it simpler to build AI applications using an open source Streamlit framework on its cloud service. Snowflake Cortex, available in private preview, provides access to a serverless computing framework through which developers and data scientists can access large language models (LLMs). The first three services provided are Snowflake Document AI, Snowflake Copilot and a Universal Search tool.

Snowflake via this service is also making available Specialized Functions to provide access to AI models that can detect sentiment, extract an answer, summarize text and translate text into a selected language. Specialized functions also make use of the company’s existing investment in machine learning algorithms to provide forecasting, anomaly detection, contribution explorer and classification capabilities.

There are also General-Purpose Functions that make it possible to chat with data using a natural language interface that outputs SQL, embed vectoring capability to customize LLMs and access search tools.

The company is also previewing support for Iceberg Tables defined by the open source Apache Iceberg format to make data more accessible to a variety of processing engines, adding tools to optimize consumption of cloud resources, and extending the reach and scope of the Horizon data governance tools it makes available to include tools to visually track data lineage and monitor data quality.

Finally, Snowflake is previewing support for workloads encapsulated in containers via its Snowpark Containerized Services framework in addition to extending the scope of the DevOps workflows it provides access to via its managed service to include a database change management capability. Snowflake has also added Snowpark support for Notebooks to write code, an application programming interface (API) for invoking machine learning models and a model registry.

Organizations will not be able to have a successful AI strategy without a strong data foundation, says Christian Kleinerman, senior vice president of product for Snowflake. “All of this sets the foundation for customers to have a strong data foundation,” he says.

It’s not clear to what degree the Snowflake platform will be employed to host applications, but the company is clearly betting the volume of data residing in its cloud platform will attract a wide range of application developers as it becomes more apparent AI will be pervasively applied across almost every application.

Of course, Snowflake is not the only provider of a data warehouse or data lake that has similar ambitions. Regardless of platform, however, the one thing that is clear is that a convergence of workflows is now underway to enable IT organizations to streamline the management of the massive amounts of data that now reside in the cloud.