Snowflake, the AI Data Cloud company, at Snowflake Expedition 2026 announced a set of innovations that make Snowflake the foundation of the agentic enterprise. Now organizations can put AI agents to work that understand the business, operate within its controls, and can be trusted to deliver real results.
These product advancements include Business Ontology,4 which grounds agents in definitions the business has approved, while Cortex AI Gateway2 governs every model, tool, and action those agents touch. With that foundation in place, Snowflake CoCo, the coding agent for data teams, and Snowflake CoWork, the personal work agent for knowledge workers, take on the recurring work that slows teams down across the business to drive impact. Pushing autonomous work even further, new CoCo and CoWork Bots5 serve as always-on AI coworkers that continuously work in the background, act directly in the tools teams already use, and check in only when a decision or sign-off is required.
Most enterprises don’t have trouble building an AI agent. They have trouble trusting one enough to put it into production. According to a recent MIT Technology Review report sponsored by Snowflake, 55% of executives cite a lack of business context as the primary cause of low-quality AI agent insights, with fragmented or inaccessible data ranking second at 45%.
An agent that doesn’t know how the company defines revenue gives a confident wrong answer that someone has to catch. Meanwhile, agents are connecting to models, MCP servers, and enterprise systems faster than IT and security can keep up, which brings new risks and unexpected bills. As a result, a lot of companies are keeping their AI agents in pilot phase, grappling with how to fully trust and control these agents in order to scale them across the business.
Snowflake is uniquely positioned to close that gap. For over 14,500 global enterprises, Snowflake already sits at the center of the data, business context, AI models, and workflows that agents depend on, serving as the governed control plane for the agentic enterprise. Today’s announcements turn that foundation into a complete system for the agentic enterprise. Context, controls, and execution are built together, so each makes the others stronger.
“The enterprises that win won’t be the ones with the most agents. They’ll be the ones whose agents know exactly what the business means by accessing the right data, abide by governance and guardrails specified, and can account for actions taken by agents,” said Christian Kleinerman, EVP of Product, Snowflake. “That’s the difference between an agent that guesses and an agent the business can run on. Snowflake already holds the data, context, and governance enterprises rely on. By extending Snowflake Horizon Catalog with Business Ontology and Cortex AI Gateway, context and governance stop being afterthoughts and become the foundation every agent is built on for the agentic enterprise.”
It Starts with Context: Agents That Know What the Business Means
Consider a sales leader who asks an agent for a weekly net revenue report. A dashboard gives one number, an analyst another, and an agent a third. For the report to be worth anything, the agent first needs to know the company’s official definition of net revenue and who signed off on it.
Business Ontology, the newest capability in Horizon Context, the home for governed business context inside Snowflake Horizon Catalog, provides that. A Business Ontology governs the official meaning, a Semantic View calculates it, and Cortex Sense infers what isn’t already defined. Together, this unified framework establishes one single, authoritative business context for every agent, app, and person across the enterprise. Teams approve official definitions for core concepts like net revenue, assign each one an accountable owner, and link it to the underlying data, inside and outside of Snowflake. However, approved definitions always take precedence. Organizations can migrate existing definitions from a third-party catalog or spreadsheet in a single guided CoCo session, and every change is logged. When an auditor asks what net revenue meant in Q2, the answer is a simple lookup.
“The value of a data product depends on everyone agreeing what each attribute means, whether it’s a household income band or a B2B firmographic segment,” said Joe Tobey, Head of Data Product Engineering, dentsu. “As agents become how people query that data, those definitions can’t live in a static data dictionary. Business Ontology gives every definition an accountable owner, links it to the data that implements it, and makes it the answer every agent returns. That’s what lets us put agents in front of our data with confidence.”
Then Control: One Set of Rules for Every Agent
An agent that knows the right definition still needs the right controls on how to access data, call models, and use tools. Cortex AI Gateway2 is the single front door for agents in an organization. As a part of Horizon Catalog, Cortex AI Gateway applies one set of policies to agent connections across clouds, frameworks, and AI providers. Agents take action on enterprise data through MCP servers, and Cortex AI Gateway applies the company’s access policies, authentication, and permissions to those calls. Agents inherit the controls already established for human users, removing the overhead of creating and managing a different set of operations.
Back to the net revenue report. Through Cortex AI Gateway Tools,3 every tool call the agent makes is checked against fine-grained authorization policies before it runs. Cortex AI Guardrails3 protect it against prompt injection and jailbreaks. Cost governance2 attributes the spend to the team that owns the report and enforces hard limits, while Dynamic Model Routing5 sends routine work to lighter, lower-cost models. Observability2 records every step in the customer’s own Snowflake account, where teams can analyze it with Agent Trace Intelligence.5 Agent Abuse Detection5 analyzes complete agent trajectories, not individual prompts, to catch agents operating beyond their permissions.
“We run a lot of AI against a lot of data, and the question we couldn’t answer was which teams and which agents were driving the spend,” said Andy Owens, VP of Data and AI, Kargo. “Cortex AI Gateway gave us that breakdown, plus hard limits so a runaway process doesn’t quietly burn through a budget. Routing each request to the right model means our routine work isn’t running on our most expensive one. We’re running more agents than ever and we know what each of them costs.”
Then Execution: Agents That Keep Working After You Log Off
With context and controls in place, enterprises can do what they’ve been holding back on: let agents work on their own. CoCo and CoWork now take on recurring work for both the data teams who build on Snowflake and the business teams who rely on it.
With CoCo Automations,2 data teams describe a workflow once and CoCo runs it autonomously at any frequency. Teams can run agents on a schedule to perform recurring or long-running tasks, like a routine pipeline health check or a weekly report someone used to build by hand. Every run keeps an audit trail and task history, so teams can see exactly what CoCo did and suspend or rerun a job whenever they need to.
CoCo and CoWork also work where people already do, through Snowflake for Slack3 and CoCo for Excel and Sheets3. Additionally, Skills and Plugin Sharing2 and Profiles in CoCo3 and new Profiles in CoWork3 mean that what one team builds, the whole company can reuse. This delivers a personalized, ready-to-use experience while helping teams standardize their work, preserving user flexibility in CoCo and central control in CoWork.
CoCo and CoWork Bots take autonomous work on the go even further. AI agents have always waited for a specific question or a predefined task before delivering value. CoCo and CoWork Bots change that dynamic entirely. Operating like an autonomous chief of staff, Bots act on behalf of the user, inheriting their exact access rights, role context, and permissions. When users set a high level goal, such as preparing briefings for every customer meeting or keeping data pipelines healthy, Bots take full initiative. They autonomously plan and complete the work in the background, operating strictly within the user’s governed boundaries in Snowflake and surfacing only when a critical decision or final approval is needed. This transforms AI from software you query into a trusted partner you empower, giving every employee continuous leverage without ever compromising control.
“It would take a lot of time to create semantic layers in the past, but things have changed drastically in the last three months since we started using Snowflake CoCo. Additionally, every piece of data flows live into Snowflake CoWork,” said Vibhor Gupta, Vice President of Enterprise Data & AI, Shelter Mutual Insurance Company. “We now have access to real-time insights on demand directly in our operational processes. With eight AI agents in production and more underway, we’ve changed how data drives decisions from the claims desk to the C-suite.”
