While much of the buzz about AI today revolves around flashy copilots and productivity hacks, the reality for most data scientists and data engineering teams remains far less glamorous. Even in 2025, they still spend much of their time on the most tedious part of the job: cleaning and preparing data, i.e., dealing with missing values, duplicates, and inconsistencies.
But Snowflake CEO Sridhar Ramaswamy wants to change that—not by replacing the people doing the work, but by eliminating the friction that slows them down, such as the endless cycles of reactive reporting. His bold bet is on agentic AI: autonomous model instances that can ingest data, reason over it, and make real-time decisions with minimal human engineering input.
“Until now, AI tools have been excellent at one-step tasks: You ask a question, you get an answer; you ask for code, you get a snippet. They are powerful assistants, but they require constant direction,” Ramaswamy tells Fast Company. “In the enterprise [space], agentic AI means goal-directed autonomy.”
From Data Silos to Conversational Insight
Ramaswamy, a former head of Google Ads, Greylock partner, and CEO of search startup Neeva, took the reins at Snowflake in February 2024 after it acquired Neeva. He brought deep AI and search expertise, quickly realigning the company’s go-to-market strategy and accelerating AI talent infusion through acquisitions like Crunchy Data, Samooha, and Datavolo. As a result, Snowflake reported its first billion-dollar quarter in May 2025, marking a 26% year-over-year increase.
Now his vision centers on embedding intelligent agents into the very fabric of Snowflake’s platform, transforming AI from a surface-level feature to a foundational layer of enterprise computing.
Ramaswamy says agentic AI moves beyond static dashboards by using smart agents that understand business goals, pull the right data, run analyses, and deliver clear, multistep answers. The real value, he adds, comes from building these agents on solid data so they can deliver lasting results across the business. “Agents built on top of a strong data foundation will unlock tremendous value across the enterprise,” he says.
This year, Snowflake has launched a wave of agentic AI-powered product releases, including Snowflake Intelligence, the Data Science Agent, Cortex AISQL, and agent-driven apps in the Snowflake Marketplace. What was once primarily a warehouse for storing and querying data is now evolving into a full-fledged “AI data cloud.” Ramaswamy believes the real opportunity lies in making AI useful and accessible to the hundreds or thousands of people who need to make data-driven decisions every day.
Today’s enterprises are overwhelmed by complexity, Ramaswamy says, and adding more disconnected AI tools only makes things worse. Snowflake’s focus has been on simplifying access to data, something it’s refined over the past decade. “You can only trust AI outputs if you trust the data foundation,” he says.
The End of Data Science? Not Quite.
Data science is fundamentally about turning raw, structured, and unstructured data into actionable insights. For years, data scientists have often been buried in technical tasks, far removed from the boardroom or customer conversation. However, with Snowflake’s Data Science Agent (currently in private preview), much of that manual effort is being automated, freeing up data teams to focus more on strategy, insight, and impact.
The agent handles data quality assessment, automatic preprocessing, feature design based on best practices, model selection and training using Snowpark code, and performance evaluation, all in under an hour. Compared with traditional workflows that take days or weeks, this dramatically accelerates pipeline creation. Moreover, generated pipelines include validated code, model lineage tracking, and integrated documentation.
According to Ramaswamy, the divide between technical and business teams wasn’t due to unwillingness to collaborate, but rather a lack of shared tools and language. Now, with AI enabling natural language as the interface to data, more people can contribute to data-driven outcomes. When experts in different fields can access insights on their own, it improves collaboration and speeds up smarter decision-making. Or, as Ramaswamy explains: “It’s about bringing the data to everyone.”
And it’s not just for data scientists or engineers. With the public release of Cortex AISQL in June 2025, Snowflake has extended SQL for data teams and business users alike with AI-native operators. Ramaswamy says enterprises’ most valuable insights have long been trapped at the intersection of structured and unstructured data, but it was nearly impossible to analyze them together. Cortex AISQL changes that by empowering data teams to query all data types.
“The distinction between a database table and a PDF will become irrelevant to the end user. You will simply ask your business question, and the platform will be intelligent enough to find and synthesize the answer from all of your enterprise data, wherever it resides,” he says.
New data operators allow users to filter, classify, summarize, and analyze text and images directly within SQL queries. Likewise, the platform’s new FILE data type can store multimedia content inside Snowflake, making it possible to work with documents, audio, images, and text alongside structured data.
For instance, a product analytics team can join sales figures with sentiment from support transcripts or defect images in a single pass, with results that are both explainable and auditable. Ramaswamy claims Cortex AI has already become a foundational pillar of many customers’ enterprise AI strategies.
For example, health wearables company Whoop used Snowflake Cortex AI to create an agent-powered chat app that makes data accessible across the organization. This tool frees up the analytics team to focus on higher-impact work (like strategy and forecasting) instead of routine data pulls.
Likewise, SaaS platform for financial services TS Imagine used Snowflake Cortex AI to build “Taia,” an AI agent that automates customer casings, work that once involved three full-time employees. Built by data analysts with little AI experience, “Taia now handles over 60,000 inquiries annually, freeing up staff for higher-value decisions,” Ramaswamy says.
The most visible piece of Snowflake’s agentic evolution is Snowflake Intelligence, a new conversational AI experience built atop LLMs from OpenAI and Anthropic. Snowflake Intelligence can handle query generation, data synthesis, and insight summarization across structured and unstructured formats.
But if AI agents can handle the grunt work of data science, where does that leave the data team?
Ramaswamy, for his part, acknowledges concerns about AI replacing jobs but believes agentic AI will actually increase the value of data scientists. He contends that instead of taking over their roles, AI will enhance them, shifting their focus from execution to strategy. “They will spend less time on the ‘how’ and more time on the ‘why’ and ‘what if,’” he says.
“We call ourselves the ‘AI data cloud’”
Snowflake’s agentic AI strategy is focused on enabling faster decisions, more scalable insights, and truly collaborative intelligence between people and machines. Ramaswamy sees this as the next evolution of enterprise tech, where data isn’t just mined, but becomes an active participant in decision-making. The company’s multiyear AI partnership with Anthropic further advances its agentic ambitions, allowing customers to leverage advanced models like Claude securely, with built-in visualizations and analyzer functionality.
If Snowflake’s bet on agentic AI pays off, it could reshape not only how companies use data but also who gets to use it, and what they can do with it.
“We call ourselves the ‘AI data cloud’ because we believe that bringing analytics, predictive, and AI capabilities to a cloud centered around high-quality data is what it takes to unlock enterprise value,” Ramaswamy says. “Snowflake’s role is to be the central brain and nervous system of the AI-native enterprise.”
Zaloguj się, aby dodać komentarz
Inne posty w tej grupie
A cryptocurrency working group formed by President Donald Trump is set to release a report on Wednesday that is expected to outline t
Google is indexing conversations with ChatGPT that users have sent to friends, families, or colleagues—turning private exchanges intended for small groups into search results visible to millions.
It’s hot. Everyone is sweating, and anyone who chooses to venture into the world armed with nothing but natural deodorant knows they’re playing a risky game.
But online, the backlash aga
If you’ve been thanking the heavens for your A/C this week, spare a thought for Paul Farmer, who’s enduring the peak of Arizona’s summer without it—by choice.
Last year, Farmer went with
Japanese electronics and technology company Panasonic has chosen a new chief executive after eking out a 1.2% rise in its first quarter
