“Integrating artificial intelligence (AI) and data into a single platform is our important goal.”
Benoit Dageville, co-founder and institutional sector chief of Snowflake, held a press conference in the Four Seasons Hotel in Jongno-gu, Seoul, on the 25th and noted this. Dageville said, “Snowflake started out as a cloud data platform but has now evolved into an AI data platform, which is a natural direction,” and explained, “Integrating unstructured data such as video, audio files, image files, and PDF documents with structured data (data existing in table form) is difficult, and AI enables this.
Snowflake is a global cloud data platform company with 11,200 customers. Among the Forbes Global 2000 companies, 745 are Snowflake customers, and the number of daily data operations exceeds 6.4 billion. At last November’s annual developer conference 'BUILD 2024', Snowflake unveiled 'Snowflake Intelligence', a data agent available in natural language. The feature allows anyone to easily access data and use the data platform using natural language, even without knowledge of conversational computer language (CICL). Snowflake also launched the AI agent 'Kortek Agent' in February, which supports operations by accessing structured and unstructured data.
Dageville emphasized the importance of building a data platform that can be used in natural language, in addition to data integration. He said, “Because only users who know conversational computer language could utilize the data platform for work, there was a significant difference in data access and usage within corporations.” He added, “Now, with Snowflake Intelligence, everyone within the corporation can access data and carry out work by asking questions in natural language and obtaining data-driven answers.”
Dageville stressed that Snowflake has enabled various large language models (LLMs) to be used within its platform without sending data externally for natural language processing. He noted, “Since we utilize data within a single platform without sending it outside, we have been cautious from a security standpoint,” and said, “Users can choose different LLMs according to their work needs, and they can freely use expensive or inexpensive LLM models based on their intended purpose, making it reasonable in terms of expense.
Dageville explained that the AI agent for corporations operated by Snowflake is differentiated in terms of accuracy. He said, “While general consumers might be satisfied with AI agents that have an accuracy of around 70-80% when asking questions in natural language, corporations require precise answers based on accurate data.” He added, “Snowflake has continuously evolved its services, developing solutions in this area,” and mentioned, “A high level of data accuracy stems from stable platform operation, which requires a multi-cloud strategy, rather than a single cloud.”
On this day, Snowflake announced plans to add Microsoft (MS) Azure-based services in its Seoul region (a collection of geographically separated data centers). Choi Gi-young, head of Snowflake's Korea branch, said, “Based on the multi-cloud strategy, we plan to offer MS Azure-based services in the Seoul region, in addition to Amazon Web Services (AWS),” and added, “We expect to create synergies in an integrated environment, including Power Business Intelligence and Azure AI.”