Snowflake Summit 2024 took San Francisco by snowstorm last week. Moscone Center was wrapped inside and out with Snowflake’s sky blue and white banners and packed with conference goers. Snowflake delivered an avalanche of announcements about new features and a clear mission to make the AI Data Cloud efficient, easy, and trusted. Customers, vendors, and partners showcased a staggering variety of real-world solutions using the Snowflake platform and ecosystem. Here are some of my key takeaways from this year’s conference.
1. GenAI made easy
It was no surprise that GenAI was the main attraction this year. Snowflake is making LLMs easier than ever to access and implement through Cortex AI. Available now are functions enabling developers to access a variety of LLMs directly in SQL, unlocking effortless access to text summarization, sentiment analysis, and other natural language use cases.
Even more exciting was the preview of Cortex AI Studio, a no-code environment in Snowsight enabling users with little or no technical skills to interact with Snowflake data using LLM technology. The demonstration during the Platform Keynote brought a member of the audience up to create a chatbot to ask questions about data in Snowflake. Not only was she able to build a chatbot in a few minutes, but also fine tune the LLM to provide more relevant answers for the use case.
Cortex AI is a game changer that will drive widespread innovation and adoption of practical GenAI use cases that can transform data into real business value.
2. Native apps bring the code to the data
At Snowflake Summit, Atrium was at the head of the pack on apps. Conference attendees loved our AI-enabled chatbot for marketers to analyze data in Snowflake using a natural language interface. But what if we could embed this app directly into the Snowflake platform? Snowpark Container Services (SCS) is the answer.
A big theme at the conference was the need for organizations to bring the code to the data. Developers at the conference were thrilled with this new capability currently in public preview. With SCS as a fully managed container offering, whatever you can build, you can run on the Snowflake platform. These apps can be deployed to users in your organization or monetized on Snowflake Marketplace.
We can’t wait to upgrade our chatbot with the latest Cortex features and deploy it internally for our marketing team. Our data engineers can’t wait to bring dbt into SCS to consolidate the data transformation pipelines into Snowflake for our clients.
3. Data scientists finally get notebooks and pandas
Snowflake continues to battle for the hearts and minds of data scientists. For me, the most significant advancement on that front is the arrival of the interactive notebook. Every data scientist I know starts their workflow with exploratory data analysis (EDA) in a notebook like Jupyter or Databricks. This feature makes the transition to Snowpark that much easier.
In addition, Snowflake now provides a pandas API to meet Python developers where they are. Existing scripts and solutions are easily converted with a few lines of code to import the API. From there, developers work with pandas as they normally would, only better because they have the full capabilities of Snowflake to work with large data sets.
4. Snowflake bursts onto the open standards scene with Iceberg support
When Snowflake CEO Sridhar Ramaswamy announced the introduction of Iceberg tables, he had to pause for the eruption of cheers from the audience. Iceberg tables are an open source format developed by Apache and widely accepted as the standard open format for analytic data. More than any other, this feature backs up Snowflake’s pledge to support accessibility and interoperability in the data landscape. Companies using the Iceberg table format have lower exposure to vendor lock-in and more access to a flexible, best-in-class approach to technology.
Support for Iceberg tables makes it significantly easier for organizations with deep investments in Iceberg to migrate to the Snowflake platform or even add Snowflake as an option in an environment with other query engines such as Spark, Flink, or PyIceberg.
In a bold move, Snowflake also announced the release of Polaris Catalog. This open source catalog for Apache Iceberg builds on the Iceberg catalog standard and enables interoperability across all the major cloud data platforms. In other words, whether you are using Spark on AWS, PyIceberg on Azure, or Snowflake on GCP, you can access a shared repository of Iceberg data. This signals Snowflake’s confidence that they are the best choice of query engine in the marketplace.
5. Everything hinges on data quality
As I watched session after session on AI-enabled data exploration, activation through native apps, and streamlining data security, it became clear that the lynchpin to all of this is data quality. Your RAG-enabled AI chatbot is only as accurate as the underlying data. Cortex Analyst, a natural language chatbot for structured data, requires descriptive metadata to give the LLM semantic context.
Recognizing this, Snowflake rolled out Horizon, a suite of tools for data management and governance to drive quality and trust. Data governors and stewards can set and implement policies for data security, compliance, and privacy. Data analysts and developers can quickly find relevant data for collaboration on apps and models to deliver insights.
Data governance is an ongoing challenge in every organization. As Snowflake brings down barriers to data access and activation, data teams and business stakeholders will have to invest in effective governance and data trust to ride the avalanche of new Snowflake features without wiping out.
6. Atrium’s happy hour packed the house
In partnership with Fivetran and Tableau, our happy hour at Southside Spirit House had a packed house. The event brought together customers, vendors, partners, and industry experts to discuss the ramifications of the conference over cool cocktails and tasty appetizers.
I was struck by the diversity of our guests from global corporations to scrappy start-ups. Everyone had a unique and compelling perspective on the conference and the impact it will have on the data platform market space.
The platform of choice for the generative AI revolution
Snowflake has made a decisive move with native apps and Cortex AI to be the platform of choice to provide the data and compute resources needed to build and deploy production GenAI solutions at scale. With the current or imminent release of a ton of new features they are enabling business and technical personas at every level to participate in the GenAI revolution.
They have also thrown in with the open source community not only supporting Iceberg tables, but contributing Polaris Catalog — to give everyone full interoperability across major SaaS applications and cloud platform providers.
Interested in how your organization can leverage Snowflake to make data fast, efficient and trusted? Contact us.