Snowflake Intelligence: your data knows everything, it is time to listen and act

Snowflake Intelligence: talking to all your data, not just your data warehouse.

In November 2025, Snowflake announced the general availability of Snowflake Intelligence, presented as an “enterprise intelligence agent” that lets every employee ask complex questions in natural language and get answers they can act on, directly inside Snowflake (*). 

In a few months of preview, more than 1,000 clients had already deployed over 15,000 agents in production, which gives a good sense of how much traction these uses have (*).

• Source Snowflake (1) (2)

In just a few months:

But behind “chatting with your data” sits a much deeper transformation: 

• Unifying structured and unstructured data on the same AI foundation. 

• Secure conversational access to the whole information estate. 

• Strengthening governance and the semantic layer, which is no longer confined to the data visualization tools. 

Those three axes are exactly where we position KPC and our Talk to Data. 

Those three axes are exactly where we position KPC and our Talk to Data.. 

Axis 1 – Unifying structured and unstructured data: widening the company's information base

Historically, Snowflake established itself as the reference platform for structured data (tables, metrics, facts, dimensions). With the AI Data Cloud and components such as Cortex AI, Document AI and Cortex Search, that promise has gradually extended to unstructured content.  

• It rests on unified access to structured and unstructured data, inside the same governed Snowflake environment. Including audio analysis.

• Thanks to Document AI, it becomes possible to extract information from invoices, contracts, emails, PDF reports and reconcile it with reference data (clients, contracts, products and so on).  (*)

• “Agentic” analytics capabilities (Agentic Document Analytics) let you ask a question that draws on thousands of documents, well beyond classic RAG architectures limited to a few dozen documents per query.  

* Source Snowflake (4)

What does that change in practice?

Until now, most data & BI projects stayed centred on figures (revenue, margin, NPS, volumes, lead times and so on) fed by a few core systems (ERP, CRM, business tools).

With Snowflake Intelligence, the company's information base widens:

• You no longer just look at “how many customer returns?”, but also “why those returns?”, by analyzing the text of tickets, emails or NPS surveys.

• You no longer just track the dispute rate, you explore the contract clauses that keep coming up in contested cases.

• You no longer settle for an IT backlog, you can surface the recurring themes in business requests automatically.

Worth remembering: In other words, Snowflake Intelligence completes the movement Snowflake started: making the platform not just a data warehouse, but the company's “datum”, the gathering point for all its strategic information — structured and unstructured alike. 

Axis 2 – Transforming access to data: from SQL query to dialogue - The second break with the past: the conversational interface.

Snowflake Intelligence is first of all a ready-to-use application, with a chat interface in Snowsight, which lets a business user ask questions in natural language and leave the agent to:

1. Interpret the question

2. Generate the right queries

3. Cross-reference the right sources

4. Return an answer you can understand.  
   

That interface rests on:

• Cortex AISQL and Cortex Analyst, able to turn a business question into SQL, then explain or enrich the results.

• Leading LLMs (OpenAI, Anthropic, Meta, Mistral, DeepSeek and others) run directly inside Snowflake, through Cortex AI. URL

• Cortex Agents that orchestrate the complex steps (breaking a question down, selecting the right sources, applying the right tools and so on). URL

For the business, that means:

• Less dependency on data teams for every new question. 

• The ability to explore, test hypotheses, dig into a “why” in real time. 

• A way of interacting that feels like a conversation with a colleague who knows the data inside out, rather than filing a report request.  

  Example: “Explain the drop in margin on BU X in Q3, taking account of sales discounts, product mix and logistics costs, and suggest three priority courses of action.” 

Snowflake Intelligence does not just produce a figure. It can: 

• Find the right tables and views, 

• Break the problem down, 

• Reason across the different factors, 

• Produce a reasoned summary, with visualizations where appropriate.  

Worth remembering: This transformation in access to data is not just a change of UX. It is a cultural change: data genuinely becomes available “on demand”, in the language the business speaks. 

Axis 3 – Semantic model and governance: when AI forces you to solidify your data stack

The third axis, often underestimated: Snowflake Intelligence puts the semantic layer back at the centre of the game. 

Snowflake Intelligence rests on a rich understanding: 

• of metrics (definitions of revenue, margin, churn and so on), 

• of business relationships (client, product, contract, region, channel…), 

• of existing BI contexts (Tableau, other data visualization tools, “hand-written” SQL and so on).  

Snowflake even introduces “semantic layer on autopilot” capabilities: the platform learns from the queries that run, from frequent questions and from definitions already present in your tools, and proposes enriching and adjusting the semantic model as it goes. 

What that changes is that:

• The semantic model no longer lives only in the data visualization tool (as local calculations, extracts, definitions scattered around). 

• It moves closer to the data foundation — inside Snowflake — and becomes a governance component in its own right, shared between IT, data and the business. 

• Every answer Snowflake Intelligence produces has to be traceable: you can go back to the tables, the views, the calculations and the business rules, which is key for compliance and for trust.

  
  

In other words, conversational AI forces you to make the whole data stack robust and interoperable: 

• Data quality and freshness, 

• Managing rights and roles, 

• Clear definition of the reference metrics, 

• Documentation and business alignment. 

Worth remembering: At KPC, we see Snowflake Intelligence as a stress test for the data foundation: if the answers are not reliable, that is not (only) an AI problem, it reveals structural weaknesses in governance and in the semantic model. 

What it means for organizations: it is not “plug & play” on the foundation side

On the user side, Snowflake Intelligence feels simple: you open an interface, ask a question, get an answer. On the organization side, though, the step up is real:

Clarify the priority use cases

• Which areas do you start with? 

• Which decisions genuinely need speeding up? 

• Which risks (regulatory, business, reputational) need framing? 

Consolidate the data estate 

• Ensure the quality and completeness of the data in the areas you target. 

• Reduce integration debt, silos, redundancy. 

Structure the semantic layer

• Identify the “golden” metrics and their official definitions. 

• Eliminate divergences between local versions of the truth. 

• Document the calculation rules, the aggregations, the filters. 

Frame how AI is used 

• Define user profiles and levels of autonomy. 

• Put mechanisms in place to evaluate and monitor the answers. 

• Build training and change management paths. 

Snowflake already provides native components for evaluating agents (such as Agent GPA, which automatically measures answer quality on standard datasets, with error detection reported as close to human performance). (*) 

But for an organization, the real question is: how do I bring those capabilities into an operational framework I control, aligned with my business priorities? 

That is where KPC comes in with Talk to Data. Discover our solution for putting Snowflake Intelligence in place.

* Source Snowflake (5) (6)

And now? 

Snowflake Intelligence marks an important step: the one where you can finally talk to all your data, in a governed environment, without multiplying tools or duplicating information. 

But that promise only fully materializes if: 

• the data foundation is solid, 

• the semantic layer is structured, 

• the use cases are prioritized and framed, 

• the users are supported. 

As a Snowflake partner and a participant in the Snowflake Intelligence Global Availability launch programme, KPC is positioned to help organizations take that step, by combining: 

• the power of the AI Data Cloud, 

• the maturity of Snowflake's agentic capabilities, 

• and Talk to Data expertise, to turn it into a genuine lever for transformation.

If you would like to explore how Snowflake Intelligence could fit into your data stack, your business use cases and your governance, we would be glad to talk. 

"There it is, data is finally free. Ask a question in natural language, get an answer you can act on, decide faster: that is already a reality. "

Article written by Mickael Kuentz, - Data & AI Director at KPC.