Share this on:
What You'll Learn
Key takeaways
- Snowflake CoWork is the current name for Snowflake Intelligence, with expanded agentic capabilities.
- It lets business users ask enterprise data questions in plain English without writing SQL.
- Cortex Analyst and Cortex Search can work across structured and unstructured data.
- A governed data foundation—master data, semantic definitions, lineage and access controls—is critical for trusted answers.
Snowflake CoWork lets someone ask a question in plain English, “how are Q4 sales trending in North America,” and get back a governed answer with the reasoning behind it. No SQL. No dashboard request. No ticket sitting in a data team’s queue. It was called Snowflake Intelligence until Summit 2026, when Snowflake renamed and substantially expanded it. This blog delves into what it does, how it works underneath the interface, and what has to be true about your data before it’s worth rolling out past a pilot.
Is Snowflake CoWork the same as Snowflake Intelligence?
The answer is yes. Snowflake CoWork is the current name for the product Snowflake launched as Snowflake Intelligence in November 2025. Snowflake renamed it at Summit 2026, and the new name reflects real product growth. CoWork can now act on data proactively instead of only answering questions, and it ships with capabilities like Artifacts, Cortex Sense, and Deep Research that Snowflake Intelligence didn’t originally have.
How does Snowflake CoWork work?
As per Snowflake’s documentation, a question typed into CoWork routes to the Cortex Agent API, the engine running underneath the conversational interface. The agent reasons through the request and calls whichever tools it needs:
Cortex Analyst
to translate the question into governed SQL against your semantic views
Cortex Search
To pull the relevant passage if the answer lives in a document instead of a table, and any other tools it’s been given access to.
Natural-language answer
synthesizes what comes back into a natural-language answer that carries its own explanation.
That routing is what lets one question span structured and unstructured data at once. A number sitting in a sales table and a clause in a signed contract are answered by two different mechanisms. Cortex Analyst for one and Cortex Search for the other. But the person asking never has to know which is which.
Example
“Summarize this customer’s complaint history and check it against their order records” pulls from a document index and a database table in the same response.
What can you do with Snowflake CoWork?
The most direct use case is self-service analytics for people who’ve never written SQL and never plan to. For instance, a sales director asking “show me win rates by region and deal size for Q3” gets a chart and an explanation of how it was calculated in seconds. The sales director doesn’t have to file a request and wait for an analyst to get to it.
The more interesting use cases combine structured and unstructured reasoning in a single step. Point CoWork at a batch of vendor contracts and ask it to flag which ones have auto-renewal clauses expiring inside 60 days. Then cross-reference those vendors against current spend to prioritize renegotiation. That’s a task that used to eat a paralegal’s afternoon and an analyst’s spreadsheet. Because CoWork’s document intelligence functions parse PDFs into structured, extractable fields, that combined query runs without anyone building a custom pipeline first.
Who is Snowflake CoWork for?
CoWork is built for two very different audiences.
1. Business Users
The finance lead, the supply chain planner, the marketing manager, get a way to ask real questions of enterprise data without waiting on someone else to build the report. That’s the headline use case, and it’s the one that gets the demo time.
2. Data Teams
Data teams get fewer routine requests. Every question CoWork answers on its own is a question an analyst didn’t have to stop and pull data for, which frees that analyst to spend time on harder problems, new data products, deeper modeling, and actual analysis
What does Snowflake CoWork need underneath it to work?
This is the part that doesn’t show up in the demo. CoWork is good when the semantic views, lineage, and governed definitions it’s built on are in place. If “customer” means five different things across five different systems, CoWork doesn’t resolve that ambiguity. It inherits it, and it answers confidently using whichever definition happens to be closest to the question asked. An agent that’s fast and wrong is a worse outcome for a data team than a dashboard that’s slow and right, because the agent’s confidence hides the error instead of surfacing it.
The real prerequisite is a data foundation.
The real prerequisite for CoWork is not a Snowflake license. It’s a data foundation. Master data that resolves to one governed record. A documented semantic layer that defines what “revenue” or “active customer” means. And lineage a compliance team can trace back to source when someone asks how an answer was generated
How do you get your data ready for CoWork?
Customer, product, and account records should resolve to a single governed identity, and not five near-duplicates spread across systems.
Core business metrics need clear, agreed-upon definitions that live in Snowflake’s semantic views.
When CoWork produces an answer, someone should be able to trace it back to the source tables and transformations that produced it, especially in regulated industries.
Role-based access control needs to reflect who should see what, since CoWork surfaces data to anyone asking the right question, not just people who used to know where to look for it.
None of this needs to be perfect before piloting CoWork on a narrow use case. But scaling it past a handful of users without this foundation in place is how organizations end up with an agent nobody trusts.
How LumenData gets enterprises ready for Snowflake CoWork
LumenData has spent years building for organizations moving onto Snowflake. As a Snowflake Premier Services Partner with more than 75 certifications across the team, our work centers on getting master data, semantic definitions, and lineage right before an organization layers agentic AI on top.
That means:
- Starting with an assessment of where data currently lives and how consistently it's defined.
- Consolidating master records so "customer" or "product" resolves to one governed identity.
- Building the semantic layer and lineage tracking that let an agent like CoWork answer confidently and correctly instead of just confidently.
It’s the same governance-first sequencing we bring to any Snowflake engagement, whether the end goal is migration, modernization, or agent readiness.
Our broader work on data modernization with Snowflake follows the same pattern. Get the data foundation right, and let CoWork work wonders.
Want to see what a CoWork-ready data foundation requires for any organization? Explore LumenData’s Snowflake services to see how we scope that work from day one.
About LumenData
LumenData is a leading provider of Enterprise Data Management, Cloud and Analytics solutions and helps businesses handle data silos, discover their potential, and prepare for end-to-end digital transformation. Founded in 2008, the company is headquartered in Santa Clara, California, with locations in India.
With 150+ Technical and Functional Consultants, LumenData forms strong client partnerships to drive high-quality outcomes. Their work across multiple industries and with prestigious clients like Versant Health, Boston Consulting Group, FDA, Department of Labor, Kroger, Nissan, Autodesk, Bayer, Bausch & Lomb, Citibank, Credit Suisse, Cummins, Gilead, HP, Nintendo, PC Connection, Starbucks, University of Colorado, Weight Watchers, KAO, HealthEdge, Amylyx, Brinks, Clara Analytics, and Royal Caribbean Group, speaks to their capabilities.
For media inquiries, please contact: marketing@lumendata.com.
Get your enterprise data ready for Snowflake CoWork
Authors
