Snowflake introduces new features that make it easier for companies to develop and deploy reliable AI solutions. The updates include multimodal app development and expanded support for GPUs.
Snowflake now allows companies to build more complex AI applications on their own data. Soon, developers will be able to build applications that can handle multimodal input. That means chatbots can handle imagery in addition to text.
Furthermore, a new SharePoint connector makes it easier to integrate internal documents into AI apps. Snowflake also provides external data integrations that respect copyrights, allowing organizations to securely access third-party sources such as news articles and scientific publications.
Rag and batch
With Cortex AI, developers can more easily combine structured and unstructured data in a single API call. This should make implementation of Retrieval-Augmented Generation (RAG) applications easier. Also, through built-in observation functions, developers can monitor their AI models for reliability and compliance, which is important for detecting anomalies and preventing unwanted outcomes.
Snowflake is expanding its batch inference capabilities to process large data sets simultaneously. Users will now have access to a wider range of pre-trained language models and embedded models, such as Llama, and multilingual models from Voyage. In addition, Snowflake is introducing the Cortex Playground, where data analysts and developers can test and compare models.
read also
Snowflake announces Snowflake Intelligence: AI agents in the data cloud
With Serverless Fine-Tuning, companies can more accurately tune language models to their specific data. In addition, Snowflake now supports working with distributed CPUs and GPUs in Snowflake Notebooks, speeding up large machine learning projects. Users can also add easier monitoring and explainability to models, providing insight into how models reach conclusions.
Snowflake Intelligence
The announcements happen during Snowflake’s Build 2024 virtual developer conference. The improvements in AI capabilities coincide with the introduction of Snowflake Intelligence. That is a comprehensive AI platform that takes advantage of SharePoint integration, among other things, and allows developers to build AI agents relatively easily.Bottom form