[San Francisco, United States = Digital Today reporter Chi-gyu Hwang] Snowflake said on June 2 (local time) it unveiled new features for its CoCo coding agent and Datastream, a real-time streaming service, at its annual Snowflake Summit 26 conference event.

Snowflake launched its native desktop app CoCo Desktop, a mobile app and a Slackbot, expanding the environments in which users can access CoCo.

It also added a VS Code extension, a Microsoft Excel extension and a CoCo plugin for Claude Code.

CoCo has also strengthened its autonomous execution capabilities. The automation feature supports repetitive and event-driven workflow automation for continuous monitoring, verification and operational processes. Role-based access controls and audit trails apply to all automation.

Newly launched Cloud Agents securely process tasks in the cloud and return results without users needing to keep a local environment open after starting work in Snowsight. A Skill Catalog also makes it possible to share and reuse verified workflows across teams.

Christian Kleinerman (크리스티안 클레이너만), Snowflake’s senior vice president of product, said, "If developing with AI becomes as simple as explaining it in words, more people will be able to contribute to an organisation’s AI strategy, and the transition to an agentic enterprise will also accelerate."

Datastream is a fully managed streaming service based on Apache Kafka. It integrates with existing Kafka ecosystems without code changes and ingests streaming data directly into Snowflake tables. It allows real-time data and AI to be combined on a single governance platform without separate streaming infrastructure.

According to the company, using CoCo and Datastream together enables real-time pipeline creation and building AI apps based on live streaming data on a single platform.

Madi Want (매디 원트), vice president of data at Fanatics, said, "Work that used to take days to resolve pipeline issues and model data is now handled within hours." Caitlin Harperty (케이틀린 하퍼티), an executive at Thomson Reuters, said, "We reduced the time it takes from legacy system modernisation to scaling AI pipelines from several weeks to a few days."