Snowflake Strengthens Enterprise AI Strategy with No-Code Studio and Enhanced Features

Following the recent announcement of Polaris—a vendor-neutral open data catalog for Apache Iceberg—Snowflake is introducing new tools aimed at empowering enterprises to develop trusted AI-powered applications more effectively.

During his second keynote at the company's annual Data Cloud Summit, CEO Sridhar Ramaswamy unveiled several enhancements to the Cortex AI service and Snowflake ML. These upgrades simplify the processes of building, governing, and managing AI applications using data hosted on the platform. Notably, a new no-code AI & ML Studio allows enterprise users to easily create AI applications for various use-cases.

Snowflake's initiative marks another milestone in the company's effort to provide customers with the tools necessary to build robust AI applications. This strategy aligns with Ramaswamy's vision since assuming the CEO role, positioning Snowflake against established AI players like Databricks, which has been AI-centric for some time. Recently, Snowflake also launched Arctic, an open LLM designed for enterprise-specific tasks.

Enhancing AI Capabilities with Cortex and Snowflake ML

Since its inception, Snowflake has prioritized establishing itself as the preferred data infrastructure for enterprises, enhancing AI and analytics capabilities.

With the rise of generative AI applications following the launch of ChatGPT, customers sought a solution for creating these applications using their stored data. In response, Snowflake expanded its ML features and introduced Cortex, a fully managed service providing AI-building blocks, including open-source LLMs for specific business applications.

Now, these offerings are further enhanced at the Data Cloud Summit.

New Features in the Snowflake Cortex AI Ecosystem

Cortex is being enhanced with four new features: AI & ML Studio, Cortex Analyst, Cortex Search, and Cortex Guard.

- AI & ML Studio: This no-code interactive interface accelerates development by enabling users to leverage state-of-the-art large language models (LLMs). Users can test and evaluate different models to identify the best fit for their applications, also benefiting from the new Cortex Fine-Tuning service, which enhances LLM performance for personalized experiences (currently available for select Mistral AI and Meta models).

- Cortex Analyst: Based on Meta’s Llama 3 and Mistral Large models, this feature allows businesses to build applications for querying analytical data stored in Snowflake.

- Cortex Search: A fully managed text search solution for retrieval-augmented generation (RAG) chatbots and enterprise search, it combines advanced retrieval and ranking technology, enabling users to build applications that utilize both vector and text-based search capabilities.

- Cortex Guard: Utilizing Meta’s Llama Guard, this feature filters harmful content across organizational data, ensuring that AI-generated responses from chatbots are safe and reliable.

Strengthening MLOps

Once applications are deployed, it is crucial to maintain oversight of the models they utilize. Snowflake ML's MLOps capabilities address this by offering streamlined discovery, management, and governance of features, models, and metadata throughout the ML lifecycle.

While the existing Model Registry facilitates governance of all AI models for enhanced personalization and automation, the new Feature Store—currently in public preview—enables data scientists and ML engineers to create, manage, and serve consistent ML features for training and inference. Additionally, the upcoming ML Lineage feature will allow teams to trace the use of features, datasets, and models across the ML lifecycle, currently in private preview for select enterprises.

All new Cortex capabilities are also in private preview, with a public release expected soon, though specific timelines remain unclear.

Mark your calendars: the Snowflake Data Cloud Summit takes place from June 3 to June 6, 2024.