Snowflake and Anthropic bring AI agents to businesses

Snowflake and Anthropic are expanding their partnership with a $200 million agreement to bring Claude models and AI agents to businesses worldwide. The collaboration will continue for several years and reach more than 12,600 Snowflake customers.

The goal is to let companies query and analyze their data in natural language, without having to gather it manually or build a separate tool for each system. Claude will be able to work with structured information, such as tables and figures, as well as documents, images, and audio.

From asking questions to delegating tasks

The partnership goes beyond offering a chatbot inside Snowflake. The two companies want to deploy AI agents capable of carrying out multistep processes: identifying the data they need, finding it within a company's environment, analyzing it, and explaining how they reached an answer.

Snowflake says Claude achieves more than 90% accuracy on complex text-to-SQL tasks in its internal tests. Text-to-SQL means converting a question written in everyday language, such as how sales evolved by region, into a query that the system can run on a database.

The offering includes several components:

• Claude Sonnet 4.5 powers Snowflake Intelligence, the enterprise agent that answers questions about structured and unstructured data.
• Claude Opus 4.5 is available in Snowflake from the first day of its release for more demanding analysis.
• Cortex AI Functions can analyze text, images, audio, and tabular data through SQL.
• Cortex Agents helps create custom agents that combine search and reasoning.
• Horizon Catalog adds governance, security, and AI usage oversight controls.

Why it matters for businesses

The main obstacle to using AI in sectors such as banking, healthcare, or life sciences is not just model capability. It also needs to access sensitive data without taking it out of the company's controlled environment, while making it possible to review every answer.

With this integration, a wealth manager could combine a client's investments with market data and compliance rules to prepare a personalized recommendation. The agent does not automatically replace the advisor, but it can reduce the work involved in gathering information and identify relevant connections.

Snowflake also uses Claude internally. Its engineering team uses Claude Code to support software development, while a sales assistant centralizes data and lets employees ask questions in natural language.

Customers already using it

Intercom uses Claude through Snowflake Cortex AI to power its Fin AI Agent customer service agent. The company says the integration has improved conversation automation and made processes more efficient for its most demanding customers.

Simon Data, a customer data platform, uses Claude on Snowflake to find patterns and connections that might previously have gone unnoticed, while maintaining its data governance controls.

For you, the change may appear indirectly: faster responses in financial services, more context-aware automated support, or internal analysis that previously required a team to prepare reports. But the result will depend on data quality, permissions, and human oversight.

The next step will be seeing how many of these agents move from demonstrations to real, repeatable tasks. That is precisely what the partnership is targeting: having AI work inside the systems where companies already store their data, with clear limits and the ability to explain its results.