“AI feeds on data, and as a data platform, we make it easier for customers to prepare their data for AI,” said Baris Gultekin, head of AI at Snowflake. Earlier this month, the renowned AI data cloud company unveiled Cortex Agents powered by Anthropic’s Claude 3.5 Sonnet, asserting the expansion of its AI capabilities.
Cortex Agents essentially help retrieve data insights by orchestrating structured and unstructured datasets. In enterprise tech, orchestrating data means collecting, organising and managing it from various sources and allowing it to be accessible easily across multiple systems.
On the sidelines of the launch of Cortex Agents, Gultekin sat with indianexpress.com to share more insights into Snowflake’s latest offering, the ever-advancing AI, and how the agentic era is here to help companies scale their operations.
Gultekin began by explaining the role of Snowflake, an AI data platform that has emerged as one of the most trusted platforms for customers around the world. “Snowflake is a data platform that customers trust with their data. They bring their data to Snowflake to secure, govern, and analyse it, using it for a variety of purposes,” he explained, adding that the company recently changed its tagline from ‘Data Cloud’ to ‘AI Data Cloud’ highlighting the growing significance of AI.
According to him, a big part of the company’s focus, moving forward, is to integrate AI capabilities into Snowflake so customers can use AI securely and efficiently. “It all starts with trust. We have built a system that is fully governed, allowing granular access controls to ensure that only the right people can see the right types of data,” Gultekin said. Additionally, the company builds high-quality AI components, such as natural language interfaces, to interact with large datasets, making it user-friendly for analysts, data scientists, and engineers.
Cortex Agents, the next leap
AI agents are autonomous systems that can perform a wide range of tasks that can enhance business productivity. They do this by handling complex tasks, often involving multiple steps in a matter of seconds. Cortex Agents are Snowflake’s new class of AI agents that can perform an array of business tasks using structured and unstructured datasets.
Acknowledging that 2025 is indeed the year of AI agents, Gultekin shared that Snowflake customers often face challenges in ensuring that agents are highly accurate and operate in a trusted, secure and governed environment. Since most customers do not want to move their data around, the company focussed on building agents that run right next to the data within a secured environment.
Story continues below this ad
“Cortex Agents supports both structured and unstructured data, which is essential for real-world applications. For example, if you ask, “What are my top-selling products?” you need structured data, but if you ask, “Why are people buying them?” you need unstructured data. Cortex Agents allows AI to seamlessly integrate both,” he said.
How can Cortex Agents help?
Talking about the practical use cases of Cortex Agents, Gultekin shared the examples of Siemens Energy that built a research agent which could analyse half a million pages of technical documents with high accuracy. Similarly, S&P Global used Cortex Agents to analyse 1,92,000 earnings calls that include both structured tables and unstructured commentary.
“This mix of structured and unstructured data is crucial for real-world applications. Cortex Agents plugs into both Snowflake tables and external sources like Google Drive and SharePoint, processing all of it with our advanced SQL engine and unstructured data processing tools,” Gultekin said. Snowflake developed Cortex Analyst for structured data and Cortex Search for unstructured data, allowing agents to determine the best source for an answer. “Our focus is on accuracy, and our text-to-SQL accuracy is the best in the market,” he added.
When asked what sets Cortex Agents apart from others, Gultekin shared three things – data governance, quality, and ability to orchestrate across structured and unstructured data, something that is unique to Snowflake.
Story continues below this ad
If Cortex Agents can help with orchestrating data, the company’s Cortex Search has also outperformed on several benchmarks. Gultekin said that the platform’s search abilities were from Snowflake’s acquisition of Neeva and it has outperformed many of its peers.
Looking ahead
Talking about adoption of Cortex Agents by Indian enterprises, Gultekin revealed that India has been very tech-forward with a young and open-to-change market. “We’re seeing demand across sectors like finance, retail, and media,” he said.
Buoyed by the feedback to Cortex Agents, Gultekin said that when it comes to AI in warehousing, he anticipates a shift from purely procedural data handling to a more semantic data analysis. “Combining structured and unstructured data will be crucial, and democratising data access using natural language will transform how businesses operate,” he shared, while talking about future AI trends in the AI data cloud segment.
He said that he foresees significant changes in how enterprises will interact with data. When asked about his advice for organisations considering AI implementation, he emphasised on the significance of having a proper data foundation. “Many enterprises are concerned that their data is not AI ready, because AI feeds on data and large amounts of data. Securing, governing, and preparing that data is an important first step,” Gultekin said.