Snowflake Debuts Native App Framework, Container Services Document AI
Snowflake and Microsoft also announced an expanded partnership covering integration and joint go-to-market.
June 30, 2023
Snowflake
Snowflake last summer indicated it intended to accelerate its push into supporting unstructured data when it announced that it was acquiring Applica, a company that offered an AI-based text and document automation platform. Snowflake said it has developed a new large language model (LLM) platform built from Applica’s generative AI technology to build more intelligence into documents and unstructured data.
During the conference keynote, Snowflake senior VP of product Christian Kleinerman told attendees that the company added support for unstructured data in its platform two years ago.
“Unstructured data is another core pillar of our platform,” Kleinerman said. “Document AI lets you ask questions in natural language from documents that you have stored in Snowflake. I don’t know that anyone has operationalized an end-to-end pipeline where you can give feedback to the answers you’re getting from the AI and be able to fine-tune and retrain the model.”
Snowflake Container Services, which promise to let developers port applications and large language models (LLMs) into secure Docker containers, is available in private preview. Slootman described Snowpark Container Services as a significant expansion of Snowpark, the developer platform launched in 2020 for developers to securely deploy Python, Java and Scala code in Snowflake.
“This is a huge expansion of Snowpark,” Slootman said. “We can take whole applications, whole sets of services. And we can now run them inside the Snowflake governance parameter. Even though this is a very, very new thing, a lot of people are already running it very, very quickly.”
For example, supply chain management software provider BlueYonder has numerous legacy engines that still need to be made available.
“They can’t rewrite them. They just want to host them inside Snowflake, use the database engine, and just make the services available, like they did before. This really helps us swoop up any function, any application that already exists.”
Slootman said Snowpark Container Services are poised to play a key role in enabling developers to build AI and machine learning models into containers.
“It’s not a coincidence that this is also going to be the strategy,” he said. “This is how we’re going to host large language models. It’s a great strategy because we don’t have to port them, we don’t have to hack them, they’re going to go wholesale into the container, and then they can be addressed by the applications themselves. We are anticipating very, very rapid development in these areas. And as a result, we need to have a strategy to adopt that and make it available to the world.”
Snowflake senior product manager Muzz Imam and senior product marketing manager Julian Forero explained in a blog that this will expand the scope of AI/ML workloads they can build with Snowpark by giving developers programming flexibility and configurable hardware options, including GPUs.
“For example, data science teams can accelerate the execution of machine learning (ML) Python libraries used in training and inference jobs. They can also run computationally intensive generative AI, such as large language models (LLMs). App developers can build and deploy front-end user interfaces using React and other popular web development frameworks. Data engineers can run optimized logic, typically written in C/C++, in the same processing engine that runs SQL or Python DataFrame operations.”
Developers can use Snowpark Container Services with the new Snowflake Native App Framework to deploy cloud-native and complex within a customer’s Snowflake account. According to Snowflake, they can securely install and run solutions such as hosted notebooks and LLMs in a manner that protects a provider’s proprietary IP because data never leaves a specific account.
Data science teams using Hex can use Snowpark Container Services to query data with SQL, R and Python tools and to run machine learning training on GPUs. According to Snowflake, the data is secured within the user’s account.
Other launch partners include A121 Labs, Alteryx, Astronomer, Dataiku, Nvidia Reka and SAS. As reported, Nvidia earlier this week revealed a strategic partnership with Snowflake.
Snowflake released the public preview of its new Snowpark ML APIs, which the company claims will enable more efficient model development. Also, the Snowpark Model Registry, in the private preview, is designed to allow for more scalable MLOps. Snowflake also announced the imminent public preview release of Streamlit in Snowflake, a tool that will convert models into interactive apps capable of advanced streaming.
While the Snowflake Native App Framework launched on the AWS Cloud, Snowflake announced an expanded partnership with Microsoft. The latest alliance includes an agreement to enhance integration between the Snowflake Data Cloud and Microsoft’s Azure OpenAI, Microsoft Cognitive Services and Azure ML. The two companies also plan to boost their field collaboration with joint go-to-market efforts with shared customers, including State Street.
While the Snowflake Native App Framework launched on the AWS Cloud, Snowflake announced an expanded partnership with Microsoft. The latest alliance includes an agreement to enhance integration between the Snowflake Data Cloud and Microsoft’s Azure OpenAI, Microsoft Cognitive Services and Azure ML. The two companies also plan to boost their field collaboration with joint go-to-market efforts with shared customers, including State Street.
Cloud database provider Snowflake is building out its new online application marketplace with the launch of its Snowflake Native App Framework. Snowflake rolled out the public preview of the new framework in the AWS cloud at this week’s Snowflake Summit 2023 in Las Vegas.
Snowflake’s new cloud-native app framework is among numerous announcements throughout the weeklong event that included partnerships and previews by the company. Many of the announcements focused on advancing the development of AI and machine learning training into applications that can tap the Snowflake cloud database. Among them were Document AI, Snowpark Container Services and the release of various APIs.
The new Snowflake Native App Framework lets developers build applications that can scale with Snowflake’s Data Cloud and utilize its operating model. Partners will soon be able to deploy their applications built in this framework in the new Snowflake Marketplace. Snowflake said it will imminently release a public preview that lets developers add monetization capabilities.
Snowflake first revealed its plan to create the app framework a year ago, and until now, it has been in private preview. The first companies planning to host native apps in the new marketplace are Bond Brand Loyalty, Capital One Software, DTCC, Goldman Sachs, LiveRamp, Matillion, and My Data Outlet.
Snowflake Native App Framework: Using Apple’s App Store Model
Frank Slootman, Snowflake’s CEO, compared the Snowflake Marketplace to Apple’s mobile app store in his keynote at the conference.
“There is now such a thing as a Snowflake application,” Slootman said. “And we’re now building applications as Snowflake applications. We’re super excited about that because they’re using all the common services, common governance frameworks, and obviously the database engine, which is really the centerpiece to the entire platform.”
Slootman claimed that the applications will run in multiple clouds as a single implementation. Also, he predicts it will grow into a substantial ecosystem.
“These apps, there are going to be hundreds of them,” he said. “If we’re successful, there are going to be thousands of them. This is really a core part of our strategy. Apps are not just limited to application developers [and] software companies. Everybody these days is a software developer. Everybody is building the customer-facing applications, internal applications, and so on.”
The Snowflake Native App Framework was one of the multiple announcements by the company to broaden the capabilities of its widely used cloud database. A roundup of them appears in the slideshow above.
Want to contact the author directly about this story? Have ideas for a follow-up article? Email Jeffrey Schwartz or connect with him on LinkedIn. |
About the Author(s)
You May Also Like