Snowflake today announced a number of platform enhancements during the Snowday 2023 event. SVP of Product Christian Kleinerman grouped the product announcements into three categories: fortifying the data foundation, celebrating artificial intelligence (AI) successes, and enhancing the capabilities for building and scaling applications.

“The rationale for us is fairly straightforward. We do not believe that organizations will be able to have a successful AI strategy and a successful application of AI without a strong data foundation,” Kleinerman said during the pre-event media roundtable.

Snowflake SVP of AI Sridhar Ramaswamy echoed, the theme is how the vendor integrates machine learning and AI into its Data Cloud platform, while “we want to make sure that more complex applications that people will want to build on their own, say a copilot or an AI-driven application, is also possible.”

Snowflake Cortex leads the AI and LLM support in the Data Cloud

At this year’s Snowday event, Snowflake announced nearly a dozen AI-related improvements, but the introduction of the Snowflake Cortex took center stage.

Snowflake Cortex, which is in private preview, is a fully managed service designed to enable organizations to discover, analyze, and build AI apps in the Data Cloud more easily by bringing industry-leading large language models such as Meta AI’s Llama 2 mode, task-specific and advanced vector search functionality directly to the users.

These tools empower teams to fast-track their analytics and rapidly craft contextualized large language model (LLM)-powered applications, enhancing productivity and innovation.

“At its core, we are bringing advanced search as well as language models right into the heart of Snowflake with a new component that we call Snowflake Cortex,” Ramaswamy said. “This brings in advanced LLMs as well as search functionality, that's based on hybrid search, a combination of directed indexing, as well as information retrieval indexing right into Snowflake.”

The data lake vendor also built three new LLM-powered experiences on Cortex, including Document AI, Snowflake Copilot and Universal Search.

  • Document AI (private preview) assists enterprises in extracting content from documents and fine-tuning results using a visual interface and natural language. “​​It makes it easy for an analyst without any specialized knowledge of programming or language models to be able to extract the structured values from these documents and put them into a table. We also make it really easy for them to set a set of a process by which as new documents come, data is automatically extracted,” Ramaswamy said.
  • Snowflake Copilot (private preview) introduces generative AI to everyday coding tasks with natural language, enabling users to interact with their data using plain text, write SQL queries against relevant data sets, refine queries and filter down insights.
  • Universal Search (private preview) allows users to find and start getting value from the most relevant data and apps for their use cases, which includes search across a customer’s Snowflake account, such as databases, views and Iceberg Tables. The function is based on the technology from the vendor’s recent Neeva acquisitions, according to Ramaswamy.
Accelerating ML model building in Data Cloud

Snowflake is also bolstering the support developers to build machine learning (ML) models and full-stack apps within the Data Cloud, including the strengthening of Snowflake’s Python capabilities through Snowpark, augmented DevOps support and containerized workload management.

  • Snowflake Notebooks (private preview) offers an interactive, cell-based programming development interface for Python and SQL users to explore, process, and experiment with data in Snowpark.
  • Snowpark ML Modeling API  (general availability soon) scales out feature engineering and simplifies model training for faster and more intuitive model development, and supports popular AI and ML frameworks natively in its platform. “One of the problems or challenges that organizations often face is so many of the tools and the tool sets that have been built for ML really struggle to work at scale and high performance,“ Kleinerman said. “Snowpark modeling APIs really make this a seamless experience for those data scientists building these workloads.”
  • Snowpark Model Registry (public preview soon) builds on a native Snowflake model entity that enables scalable and secure deployment and management of models, such as deep learning models and open-source LLMs from Hugging Face. “With the model registry, Snowflake is now a first-class source for you to take that model that you've trained, whether you trained it in Snowflake or even if you trained it outside of Snowflake, but you can register that model with Snowflake next to your data,” Kleinerman said.
Snowflake eliminates data silos, extends governance capabilities

Snowflake also introduced a suite of advancements aimed at consolidating data, enhancing governance and compliance and increasing cost transparency.

The data lake provider enhanced its support for Iceberg Tables, a new table type it introduced last year, which ensures users can seamlessly unite all of their data together in the Data Cloud.

Additionally, the vendor enhanced Snowflake Horizon, which is a built-in governance solution that unifies its compliance, security, privacy, interoperability, and access capabilities in the Data Cloud. Snowflake added several new capabilities including additional authorizations and certifications, data quality monitoring, data lineage UI, differential privacy policies, enhanced classification of data and a trust center.

To help customers better optimize spend, Snowflake is private previewing the new cost management interface that acts as a single place for admins to easily understand, control and optimize their spending with out-of-the-box capabilities.

Investing up to $100M in innovative apps

Snowflake launched its Powered by Snowflake Funding Program at its Snowday 2023 event, which plans to invest up to $100 million to support early-stage startups building Snowflake native apps.

The program features venture capital firms such as Altimeter Capital, Amplify Partners, Anthos Capital, Coatue, ICONIQ Growth, IVP, Madrona, Menlo Ventures, Redpoint Ventures and Snowflake Ventures.

Meanwhile, as part of the program, Amazon Web Services (AWS) will further provide up to $1 million in Snowflake credits on AWS over four years to startups building Snowflake native apps.

Image credit: Snowflake Inc.