{"id":14075,"date":"2026-07-27T20:03:26","date_gmt":"2026-07-27T20:03:26","guid":{"rendered":"https:\/\/cheesecakelabs.com\/blog\/"},"modified":"2026-07-27T20:03:26","modified_gmt":"2026-07-27T20:03:26","slug":"agentic-analytics-production","status":"publish","type":"post","link":"https:\/\/cheesecakelabs.com\/blog\/agentic-analytics-production\/","title":{"rendered":"Agentic Analytics in Production: What Snowflake CoWork Replaces, and What It Doesn’t"},"content":{"rendered":"\n

For twenty years, the answer to “can I get a view of X?” has been the same: someone opens a ticket, a data team builds a dashboard, and three weeks later the business question has changed. That model is ending.<\/p>\n\n\n\n

Every major data platform now ships a conversational, agent-driven layer on top of the warehouse \u2014 Snowflake CoWork, Databricks Genie One, Microsoft Fabric Data Agents \u2014 and the architecture behind all three is converging on the same three parts.<\/p>\n\n\n\n

We build these systems for a living. This post covers what we’ve learned shipping them to production \u2014 mostly on Snowflake, where the tooling matured fastest, but the architecture translates across all three platforms.<\/p>\n\n\n\n

What is agentic analytics, and how does it work?<\/h2>\n\n\n\n

The pattern has three parts, whatever the vendor calls them:<\/p>\n\n\n\n

The semantic layer defines what’s true<\/h3>\n\n\n\n

A governed definition of your business \u2014 entities, metrics, joins, and rules like “hours are computed per weekday only” or “actuals only count approved records.” This layer is where correctness lives. Worth noting that the three platforms implement it quite differently: a Snowflake semantic view is a declarative object you author.<\/p>\n\n\n\n

Databricks’ Genie Ontology is a self-improving knowledge graph that auto-extracts meaning from tables, queries, dashboards, and connected apps; and Fabric semantic models inherit the Power BI modeling lineage<\/strong>. Same job, three philosophies \u2014 authored, learned, and BI-native.<\/p>\n\n\n\n

Agents inherit the user’s permissions<\/h3>\n\n\n\n

Purpose-built Artificial Intelligence<\/a> workers pointed at that semantic layer, each with a scoped goal: a churn agent on the sales data, a capacity agent on resourcing data. They translate natural language into SQL, run it, chart the results, and critically, run under the user’s own credentials<\/em>.<\/p>\n\n\n\n

In Snowflake’s words, “all role-based access control and data-masking policies associated with the user automatically apply.” If you can’t see a table, neither can your agent. No new security surface.<\/p>\n\n\n\n

The chat interface replaces the ticket<\/h3>\n\n\n\n

A dedicated workspace \u2014 Snowflake CoWork is now GA with a companion iOS app (Face ID, voice input, role and warehouse selection) \u2014 where a business user selects an agent and simply asks. Roles keep them scoped: they can query, but they can’t touch a warehouse or a pipeline<\/strong> by accident.<\/strong><\/p>\n\n\n

\n
\"A<\/figure>\n<\/div>\n\n\n
\n

Read more: <\/strong>When to Move Your Data Out of Spreadsheets?<\/a><\/p>\n<\/blockquote>\n\n\n\n

Where does agentic analytics pay off first?<\/h2>\n\n\n\n

The highest-ROI use cases we see share one trait: recurring questions with changing parameters, exactly what dashboards handle worst.<\/p>\n\n\n\n

Which accounts are churning, and what do they share?<\/h3>\n\n\n\n

“Which accounts dropped usage more than 20% this quarter<\/strong>, and what did they have in common?” A churn agent over the sales semantic layer answers in seconds \u2014 and the follow-up question (“break that down by plan”), which would normally mean another ticket, is just the next message.<\/p>\n\n\n\n

Who has open capacity in the next two weeks?<\/h3>\n\n\n\n

Time tracking, hours-bank balances, utilization, capacity planning. “Who has open capacity in the next two weeks?” “Is anyone trending toward overtime this month?” Ops managers ask these weekly; the parameters change every time. This is one of the domains where we’ve deployed agents<\/a> in production, and the ticket-queue effect disappeared almost immediately.<\/p>\n\n\n\n

Why is this line item up 14% this month?<\/h3>\n\n\n\n

Month-end close is a burst of one-off questions \u2014 variance explanations, anomaly hunts, “why is this line up 14%?” \u2014 that traditionally lands on an analyst’s desk all at once. An agent absorbs the burst.<\/p>\n\n\n\n

What executives ask when they never open the dashboard<\/h3>\n\n\n\n

Leaders rarely open dashboards; they ask people. Giving them an agent (especially on mobile) meets that behavior instead of fighting it.<\/p>\n\n\n\n

The common thread: you’re not replacing reports, you’re absorbing the long tail<\/em> of questions that never justified building one.<\/p>\n\n\n\n

\n

Read more:<\/strong> Your AI Strategy Has a Data Problem<\/a><\/p>\n<\/blockquote>\n\n\n\n

What surprised us running agentic analytics in production?<\/h2>\n\n\n\n

A few findings that don’t show up in the marketing pages:<\/p>\n\n\n\n

Setup takes four SQL statements. The semantic layer takes months<\/h3>\n\n\n\n

Enabling CoWork and deploying a working agent takes four SQL statements. Snowflake’s own coding agent, CoCo (formerly Cortex Code), can even write the agent specification for you, because it reads the semantic view and understands the data model.<\/p>\n\n\n\n

The real engineering effort, and where experience matters, is the semantic layer itself. If your business rules aren’t mapped and described there, no model will save you<\/strong>. This is the new data modeling, and it’s a skill, not a checkbox.<\/p>\n\n\n\n

Agents catch their own mistakes<\/h3>\n\n\n\n

In one deployment, an agent noticed the totals it computed were implausibly high for a single person, traced the anomaly to duplicate rows in a directory join, corrected the query, and explained the correction<\/em> before presenting results. That self-verification loop is the difference between a text-to-SQL demo<\/strong> and something you can put in front of a stakeholder.<\/p>\n\n\n\n

Adoption follows trust, and trust follows scoping<\/h3>\n\n\n\n

Broad “ask anything about the company” agents underperform: they hedge, they miss business rules, and users stop trusting them after one wrong answer. Narrow agents<\/strong> with explicit instructions (“use approved records; prefer weekly aggregates; always chart”) are consistently right, and adoption compounds from there.<\/p>\n\n\n\n

Cost is usage-based, and worth understanding before you commit<\/h3>\n\n\n\n

Since April 2026 Snowflake bills AI in flat-priced AI Credits<\/em> \u2014 $2.00 each<\/a> (global routing; $2.20 regional) \u2014 decoupled from your Snowflake edition. But an AI Credit is not a million tokens: consumption is metered per million tokens, input and output billed separately, and the credits-per-million rate depends on the orchestration model you pick.<\/p>\n\n\n\n

A capable model can run several credits per million input tokens and multiples of that on output, so real cost per million tokens lands well above $2 \u2014 check the Service Consumption Table<\/a> for your model. Add ordinary warehouse compute (attributable per agent) on top. The headline that still holds: there’s no per-seat licensing, and a scoped pilot is genuinely cheap \u2014 but “roughly $2” is the credit price, not your token bill.<\/p>\n\n\n\n

It fills a real gap<\/h3>\n\n\n\n

Snowflake has no native analytics layer for non-technical users. If your stack is Snowflake without a BI tool on top, or with one that only the analysts open, this is the shortest path to genuine self-service. Business users don’t even need to see the Snowflake console: CoWork is its own link, scoped by role.<\/p>\n\n\n\n

\n

Read more: <\/strong>Spec-Driven Development: How to Capture Intent Before You Burn Tokens<\/a><\/p>\n<\/blockquote>\n\n\n\n

Does agentic analytics replace your BI tool?<\/h2>\n\n\n\n

Not yet, and the honest answer has more nuance than “no.”<\/p>\n\n\n\n

CoWork does<\/em> produce visuals, and they persist: every chart or table it generates becomes a saved Artifact<\/strong> you can pin, share, and revisit without re-asking. It supports most Vega-Lite chart types like bar, line, heatmaps, box plots, dual-axis, small multiples, so the output is real, not a throwaway image in a chat log. If your definition of “dashboard” is “a chart I can come back to,” CoWork already clears that bar.<\/p>\n\n\n\n

What it is not<\/em> is a governed BI product. It doesn’t give you a curated single-pane board of a dozen KPIs laid out just so, on a scheduled refresh, distributed by email or embedded for people who never log into Snowflake, with alerting and pixel-perfect formatting. That’s still Power BI \/ Tableau \/ Snowsight dashboard territory: curated, recurring monitoring with wide distribution and predictable cost.<\/p>\n\n\n\n

So the division of labor is: monitoring stays on glass; exploration moves to conversation.<\/strong> What agentic analytics eliminates is the dashboard built to answer one question, once, that nobody maintains afterward.<\/p>\n\n\n\n

Teams that frame it this way get adoption; teams that announce “we’re replacing Power BI” get a turf war. One caveat worth flagging to any client: this frontier is moving fast \u2014 the CoWork rebrand added workflow automation<\/strong> and multi-agent orchestration \u2014 so the “can’t build dashboards” line is likely to soften over the next few releases. Bet on the division of labor, not on the current feature gap.<\/p>\n\n\n\n

How do Snowflake CoWork, Databricks Genie One, and Fabric Data Agents compare?<\/h2>\n\n\n\n

Here’s the part that matters if you’re choosing (or already chose) a platform: the architecture is converging, so the skills<\/em> transfer.<\/p>\n\n\n\n