Agents are built for the question you're asking right now. Governed data workflows are built for the question you'll ask every month for the next five years.
What Snowflake AI agents can do today
Snowflake CoCo is a data-native coding agent that can generate, modify, improve, and explain SQL files and Python Notebooks through conversation, then preview AI-suggested changes in a diff view before you apply them. It can stand up dbt pipelines and create semantic models for Cortex Analyst from natural language. It also searches tables and columns in plain English, pulls tags and masking policies from the Horizon Catalog, and returns lineage context with its answers.
CoCo runs in three surfaces: Snowsight, a desktop IDE, and a command-line interface (CLI). It ships with role-based access control (RBAC), OS-level sandboxing, a three-tier approval system, and a SQL read-only mode.
Governance is the other half of the story. Agent Identity, in public preview since Summit 2026, gives every agent a cryptographically verified identity with scoped, expiring permissions before it touches production data. Not every capability has reached the same maturity. Cortex Sense, the context layer intended to ground agents in your organization's business definitions, remained in private preview as of Summit 2026 and had not yet reached general availability.
How Snowflake AI agents reduce the analyst backlog
The clearest win for Snowflake AI agents is on the exploratory questions that used to clog the ticket queue. An operations analyst at a freight company investigating an on-time delivery dip can draft the join across shipment and carrier tables in CoCo, inspect the suggested SQL, and iterate before sending a request to the platform team's backlog.
For exploratory work and first drafts, Snowflake AI agents deliver a productivity gain platform teams can measure.
Why recurring reporting needs more than SQL in a repo
The exploratory win runs into a different problem the moment the same question comes back next week. SQL that CoCo generates can go into version control and become a team asset. The question is whether a SQL file in a repository gives the whole team — including analysts who don't write SQL — the visibility they need to validate and maintain the logic over time.
Repeatability across reporting periods
When a month-end report depends on a SQL query in a repository, any analyst fluent in SQL can read whether the join is at the right grain and whether the filters match business requirements.
Analysts who don't write SQL cannot. A visual data workflow exposes the same logic — the join condition, the aggregation level, the filter chain — in a form the whole team can inspect, not just the people who built it. When the grain shifts between periods, the visual diff catches it before the report ships.
Change history the whole team can review
SQL in version control gives you a change record, but interpreting a SQL diff requires SQL skills. A diff that shows a join condition changed from claim_id to claimant_id is unambiguous to an engineer; to an analyst responsible for the report's accuracy, it may not be.
A visual workflow surfaces the same change as a step anyone on the team can read: a renamed join key, an added filter, a changed aggregation. The review step works when the whole team can participate, not only the data engineers.
Ownership a non-technical analyst can pick up
Any tool carries turnover risk: the analyst who built the workflow may leave, and the next person inherits it. What's different with a visual workflow is who can pick it up. A Prophecy workflow comes with AI-generated documentation that explains each step in business terms, and the visual interface lets a non-technical analyst understand what the workflow does without reading the underlying code.
A SQL file in a repository requires the next person to have the SQL skills to trace the logic. The workflow any analyst can read is the one that actually survives turnover.
How to combine Snowflake AI agents with governed data workflows
The first thing to do is route them by use case. Use Snowflake AI agents for exploration, then promote recurring work into a versioned data workflow. When an agent's output becomes a weekly report or a month-end close input, it deserves production treatment.
Platform leads who approve agent use can apply the promotion pattern to handle the governance question. Blocking agent access drives shadow spreadsheets; when companies lock employees out of sanctioned tools, employees route around the controls.
A visual, versioned workflow the whole team can read is easier to review than a prompt history, and it deploys to your Snowflake platform under your controls.
Promote Snowflake AI agent output into governed data workflows with Prophecy
Snowflake AI agents are great for exploration, but recurring reporting still needs a versioned, team-owned data workflow with named ownership and lineage regulators can inspect.
Prophecy is an AI data prep and analysis platform whose agents generate visual data workflows from natural language — workflows non-technical business analysts can read and validate step by step, not just the engineers who built them. Where CoCo generates SQL, Prophecy generates a visual workflow the whole team can inspect, so the exploration your team does in CoCo can be promoted into a governed asset everyone owns.
Prophecy gives Snowflake teams:
- Visual workflows for non-technical analysts: Prophecy's agents turn prompts into visual data workflows you can inspect step by step, so analysts can refine joins, filters, and transformations — and validate the logic — without reading SQL.
- AI-generated documentation: Every workflow ships with documentation explaining each step in business terms, so the team can maintain it even after the original analyst moves on.
- Cross-platform data workflows: Prophecy's agents build workflows that run natively on Snowflake, Databricks, and BigQuery, and connect to external data sources — so the same governed approach extends beyond a single cloud platform.
- Cloud-native on Snowflake: Workflows deploy natively to Snowflake, sign in with your Snowflake credentials, and inherit Snowflake Horizon governance, RBAC, and audit logging.
With Prophecy, your team can keep drafting in CoCo while giving recurring reporting the named ownership and lineage it needs. Book a demo to see Prophecy build governed data workflows natively on Snowflake.
Ready to see Prophecy in action?
Request a demo and we’ll walk you through how Prophecy’s AI-powered visual data pipelines and high-quality open source code empowers everyone to speed data transformation
