Snowflake Bets Big on AI agents with a 10-minute Build Promise

The AI industry is evolving fast, and the rise of AI agents makes it clear. What began as simple LLM wrappers has grown into full-fledged systems. OpenAI’s Operator demonstrates the breadth and power of these agents.

A report from Research and Markets projects the global AI agents market to grow from $5.1 billion in 2024 to $47.1 billion by 2030. This represents a compound annual growth rate (CAGR) of 44.8%.

At Cypher 2025, Sarita Priya Darshani, senior solution engineer at Snowflake, demonstrated how enterprises can build AI agents in minutes using Snowflake Intelligence, its new no-code agentic AI platform.

“I’ve built agents in under ten minutes,” Darshani said.

Tackling data chaos with AI

Darshani explained that enterprises are struggling with fragmented data, including sales tables, PDFs, Slack messages, call transcripts, and more. This makes it hard to extract insights or scale AI systems. Snowflake Intelligence unifies these sources and layers AI models on top to deliver search, analytics, and recommendations.

The platform blends structured and unstructured data to answer not just what happened but why it happened and what to do next. In a demo, Darshani showed how sales data combined with customer feedback revealed weak marketing and inventory shortages behind poor performance in western India. The system then recommended corrective actions.

Accuracy, trust, and security

“Business leaders want three things before adopting agentic AI: accuracy, trust, and security,” Darshani said. Snowflake addresses these needs with the Horizon Catalog, role-based access, and observability tools that provide transparency into AI insights.

It integrates with leading large language models (LLMs) from OpenAI, Meta, Anthropic, and Mistral. Its Cortex Analyst supports structured data queries, while Cortex Search enables retrieval-augmented generation (RAG). Through its Model Garden, enterprises can select models suited to their use case.

Snowflake has also introduced Cortex Agents, a managed service now in public preview. These agents integrate and process structured and unstructured data at scale with built-in governance. They plan tasks, select tools, execute workflows, and refine responses.

Cortex Analyst delivers 90% accuracy in text-to-SQL, while Cortex Search has shown 12% higher retrieval accuracy than OpenAI’s embedding models, she observed. The framework uses Anthropic’s Claude 3.5 Sonnet for reasoning, coding, and workflow execution within Snowflake’s secure environment.

A user can, for instance, ask for top distributors by revenue and then request contract details. Cortex Agents split and execute these tasks across the right tools.

To improve oversight, Snowflake introduced Cortex AI Observability, powered by TruLens. It uses methods like “LLM-as-a-judge” to trace and measure agent performance.

From raw data to insights

Darshani said the platform delivers 90%+ accuracy, with a “human-in-the-loop” for greater oversight when needed.

From insurance underwriting to retail order management and supplier analytics, Snowflake Intelligence aims to cut costs, reduce cycle times, and accelerate innovation.

“With Snowflake Intelligence, we’re taking enterprises from raw data to actionable insights in the shortest path possible—without coding, without hops, and without complexity,” Darshani said.

The post Snowflake Bets Big on AI agents with a 10-minute Build Promise appeared first on Analytics India Magazine.