Snowflake, the AI Data Cloud company, announced a series of new innovations and enhancements to its Snowflake Cortex AI platform at Snowflake Summit 2024. These updates aim to democratize enterprise AI by making it more accessible, effective and secure for a wide range of people. users.
The new advances are primarily aimed at improving the ease with which organizations can create AI-based applications. Notably, Snowflake introduced new chat features that enable businesses to develop chatbots that can directly interact with their corporate data in minutes. This development aims to speed up the process of getting answers and insights from business data.
Baris Gultekin, head of AI at Snowflake, highlighted the transformative potential of these updates. “Our latest advancements in Snowflake Cortex AI remove barriers to entry so all organizations can leverage AI to build powerful AI applications at scale and unlock unique differentiation with their business data in AI Data Cloud,” Gultekin said.
Snowflake has launched two new chat features: Snowflake Cortex Analyst and Snowflake Cortex Search. These tools aim to enable users to quickly create chat experiences, leveraging both structured and unstructured data. Cortex Analyst is built using templates from Meta’s Llama and Mistral, allowing businesses to securely build applications based on their analytical data. Meanwhile, Cortex Search uses advanced retrieval and ranking technology acquired from Neeva, providing hybrid search capabilities across diverse datasets.
Awinash Sinha, Corporate CIO at Zoom, attested to the impact of these new tools. “By combining the power of Snowflake Cortex AI and Streamlit, we were able to quickly build applications leveraging large pre-trained language models in just a few days. This allows our teams to use AI to quickly and easily access useful answers,” noted Sinha.
Mukesh Dubey, Product Owner Data Platform at Bayer, highlighted the additional benefits. “The heart of this feature is high-quality responses to a natural language query on structured data. This is exactly what Snowflake Cortex Analyst allows us to do. What excites me most is that we are just getting started, and we can’t wait to unlock more value with Snowflake Cortex AI,” said Dubey.
Data security and governance remains a high priority. To this end, Snowflake introduced Snowflake Cortex Guard, which integrates Meta’s Llama Guard technology to filter and report harmful content in organization data. This addition should help ensure that AI models used by businesses are safe and reliable.
Snowflake also announced new AI-powered prebuilt experiences to further improve productivity. The Document AI feature uses Snowflake’s multimodal LLM, Snowflake Arctic-TILT, to extract content from documents, even outperforming GPT-4 in the DocVQA benchmark. Companies like Northern Trust are using Document AI to process documents at scale with greater efficiency.
Additionally, the Snowflake Copilot text-to-SQL wizard combines the capabilities of Mistral Large with Snowflake’s proprietary SQL generation model. This tool aims to improve the productivity of SQL users.
To enable no-code AI development, Snowflake introduced Snowflake AI & ML Studio, which provides an interactive, no-code interface for AI development. The studio is designed to help teams quickly develop and test AI models, accelerating the move to production and optimizing operating costs. Additionally, Cortex Fine-Tuning functionality enables serverless customization, improving LLM performance to deliver more personalized experiences.
Snowflake ML, the company’s unified platform for generative AI and machine learning, now includes advanced MLOps features such as the Snowflake Model Registry and Snowflake Feature Store. These tools aim to streamline the management and governance of AI models and features throughout their lifecycle.
Snowflake Summit 2024 also saw the announcement of new tools and collaborations, including a partnership with NVIDIA to create custom AI data applications. These developments are expected to provide organizations with increased flexibility and interoperability in their data operations.
