What happened
At its Expedition 2026 conference, Snowflake released new versions of its CoCo and CoWork AI agents, introducing 'Bots' for continuous background operation, and launched Cortex Training, a service for customizing open- AI models using proprietary data.
Snowflake Inc. announced updates to its products and introduced a new model training service at its Expedition 2026 conference on November 3, 2026. The company released new versions of CoCo, an agent for data teams, and CoWork, an agent for business users. According to Investing.com, these updates include 'Bots' that operate continuously in the background and surface only when decisions are required.
The report states that CoCo is used by more than 9,100 accounts and CoWork by more than 5,800 accounts. CoCo Automations allows data teams to schedule repeated workflows with audit trails, while CoWork can run recurring analysis for business teams. The agents now integrate with Slack, Excel, and Google Sheets, and are available through a mobile application.
Snowflake also introduced Skills and Plugin Sharing, which allows users to share CoCo workflows across organizations. New Profile features in both agents provide role-specific context, with CoWork Profiles giving administrators control over agent access and routing instructions. The company announced Cortex Training, a service that enables organizations to customize open- AI models using their proprietary data.
Additionally, the company introduced the Snowflake AI Data Package, which includes more than 60 curated industry datasets with semantic views and a CoWork agent. Agentic Discovery on Snowflake Marketplace allows users to find and install data products through conversational interfaces. Investing.com notes that several features are in preview stages, with some available generally, others in public preview, and Bots scheduled for private preview.
Source details: ca.investing.com
Why it matters
This expansion positions Snowflake as a central hub for enterprise AI automation, moving beyond data storage to active agent management and model customization. The introduction of Cortex Training offers a potentially cost-effective alternative to existing serverless training options, while the new agent features aim to reduce manual oversight for data and business teams. These moves could accelerate AI adoption in enterprises by integrating autonomous workflows directly into existing data infrastructure, though the practical impact depends on the reliability of the new 'Bots' and the accuracy of the cost savings claims.
The launch of Cortex Training is significant because it targets a specific pain point in enterprise AI: the cost and complexity of customizing models. Snowflake claims the service shows up to 61% lower cost compared to existing serverless training options, based on pricing from Fireworks and Tinker published on October 22, 2026. This claim is not independently verified in the source but suggests a competitive shift in the model training market.
The introduction of 'Bots' represents a move toward more autonomous AI operations. By operating continuously in the background and only surfacing when decisions are required, these agents aim to reduce the cognitive load on data and business teams. This is a practical implication for enterprises looking to automate routine data tasks without constant human supervision.
The integration of CoCo and CoWork with popular tools like Slack, Excel, and Google Sheets lowers the barrier to entry for AI adoption. By meeting users where they already work, Snowflake is positioning its AI agents as practical, everyday tools rather than specialized software. This could drive broader adoption among non-technical business users.
The availability of the Snowflake AI Data Package with curated industry datasets and semantic views addresses the challenge of data quality and relevance in AI applications. By providing pre-curated data, Snowflake is attempting to streamline the process of building and deploying AI models, which can be time-consuming and resource-intensive.
Interactive Mechanism: How It Actually Works
Explore the underlying technology behind this development interactively.
crm_get_transaction(id='4092').An agent must create a draft calendar event for Tuesday at 2 p.m. Which evidence would establish the requested result?
What to watch next
Monitor the transition of 'Bots' from private preview to general availability, independent benchmarks of Cortex Training's cost and performance compared to Fireworks and Tinker, and customer adoption rates for the new Slack, Excel, and Google Sheets integrations.
The transition of 'Bots' from private preview to general availability will be a key indicator of Snowflake's confidence in the reliability and security of its autonomous agents. Enterprises will likely scrutinize the audit trails and decision-making processes of these Bots before widespread adoption.
Independent benchmarks of Cortex Training's cost and performance compared to Fireworks and Tinker will be crucial for validating Snowflake's 61% cost savings claim. If the claims hold up, Cortex Training could become a preferred option for organizations looking to customize open- models.
Customer adoption rates for the new Slack, Excel, and Google Sheets integrations will indicate the practical utility of CoCo and CoWork. High adoption rates would suggest that the agents are meeting the needs of data and business teams, while low adoption could signal usability or reliability issues.
The expansion of the Snowflake AI Data Package and Agentic Discovery on Snowflake Marketplace will be important to watch. The success of these features will depend on the quality and relevance of the curated datasets and the effectiveness of the conversational interfaces for finding and installing data products.