Written by Rishabh Srivastava (Lead Machine Learning Engineer) & Henry Lindsay Smith (Head of Data Solutions)
Mantel is pleased to announce its role as an official launch partner for Snowflake Intelligence.
This innovative agentic platform dramatically shifts how teams interact with data, moving beyond traditional dashboards and complex queries to enable true conversational analytics. As a launch partner, Mantel has gained early access and experience, leading us to believe this new service has the potential to significantly enhance how our clients manage their data, empowering every user to unlock rich, governed, and actionable insights directly through natural language.
Chatbots are rapidly becoming the standard interface for AI — and for good reason. They represent the most natural way humans interact: through conversation. In the context of data analytics and insights, this means that instead of navigating complex tools or dashboards, users can simply ask questions, explore data, and act on insights in plain language.
In 2024 Snowflake launched Cortex Analyst, which was a brand new capability that enabled users to ask questions of a structured dataset using natural language. We saw some real excitement around this capability and its ability to enhance the experience of dashboards. Fundamentally, Cortex Analyst allows users to ask ‘what’ questions and understand what has happened. This capability enables time-poor BI teams to service the ad-hoc questions of their business customers via channels like Teams, Slack or a Streamlit app.
Once people used Cortex Analyst, the feedback we heard the most was, ‘This is great at telling me what happened, but I want to know why it happened.’ This was not yet possible, as Cortex Analyst was based on turning natural language into SQL. Snowflake recognised this need and has been rapidly innovating to help users take the next step and move from ‘What?’ to ‘Why?’. In short, how could they combine insights from structured data (via SQL), unstructured data (Cortex Search) and public data?
Introducing Snowflake Intelligence
Snowflake Intelligence fills that gap. It extends far beyond “text to SQL” by introducing a unified agentic platform that can reason over data and deliver richer, more contextual insights. Instead of simply answering SQL-translated questions, Snowflake Intelligence agents can diagnose drivers behind trends, combine multiple datasets, and provide tailored, domain-specific guidance.
In other words, it transforms conversational analytics from a passive Q&A experience into an active, insight-generating workflow that is conversational, transparent and actionable. This means that any user can uncover value from their company data.
From a technical perspective, Snowflake Intelligence is built on Snowflake’s trusted foundation of security, governance, and scalability. This means it can empower every business function to interact with data using natural language, while engineering and data governance teams retain control of how this is done.
The three pillars of Snowflake Intelligence
Above: Snowflake architecture showing the 3 pillars and how the agents interact with different tools to converse with the user.
1. Conversational
Ask questions in natural language and explore your data naturally. Follow up on answers, share insights across agents, and even connect insights between business domains. Whether you’re using Cortex Analyst, Cortex Search, or custom-built tools, the experience remains intuitive, consistent, and seamless.
2. Transparent
Snowflake Intelligence makes every interaction explainable. Snowflake Intelligence provides its thinking and reasoning, making the audit of agents easier. Additionally, teams can embed verified content into the agent, which improves the answer being provided to the end user. Every event in an interaction is logged, ensuring clarity around how insights are generated and decisions are supported.
Let’s explore this in practice using a temperature comparison between 2 locations:
Above & below: Examples of Snowflake intelligences Conversational and Transparent reasoning about the user’s request and then using Analyst to generate sql.
3. Actionable
Beyond raw metrics, Snowflake Intelligence focuses on “why”: explaining drivers, correlations, and trends behind the numbers. Insights are designed to inspire action, transforming data from static dashboards into dynamic business decisions.
Continuing with the example from above, Snowflake Intelligence created insights into the 2 location’s temperature variances:
Key Insights from Snowflake
- Temperature gap: Brooksville consistently averages 15-25°F warmer than Mackinaw. Most notably on November 9th, when Brooksville reached 72.7°F while Mackinaw dropped to just 32°F – a 40-degree difference.
- Seasonal transitions: Both locations show typical fall cooling patterns, but Mackinaw is experiencing more dramatic temperature swings. Illinois has already seen some near-freezing temperatures (28-32°F range), while Florida remains comfortably mild (43-73°F range).
- Weather stability: Brooksville demonstrates more consistent temperatures with smaller daily variations, while Mackinaw shows greater volatility, particularly in mid-to-late October when temperatures ranged from the upper 60s to the low 40s.
- Climate zones: This data clearly illustrates the difference between Florida’s subtropical climate and Illinois’ continental climate, with Mackinaw entering winter conditions while Brooksville maintains pleasant fall weather.
Above: Examples of a chart and insights Snowflake intelligence created in response to the original query from the user about the average temperatures in Brooksville and Mackinsaw.
Below: The charts are customisable.
Opportunities unlocked by Snowflake Intelligence
- Rapid time to production: Create and deploy agents quickly using built-in security and governance.
- Unified feedback loop: Capture user feedback within SI to continuously refine insights and build purpose-built dashboards.
- Cross-application integration: Connect with external tools like Slack, Teams, or email using custom tools for automated workflows.
- Research mode: Dive deeper into complex questions and discover nuanced, multi-dimensional insights.
Trust, security, and governance – Built in
Every interaction within Snowflake Intelligence inherits Snowflake’s enterprise-grade security framework:
- Role-Based Access Control & Identity and Access Management (IAM): Snowflake Intelligence ensures that every user only sees the data they’re meant to see. Access is automatically tied to each person’s role in the organisation, and identity checks happen seamlessly in the background — keeping data secure without adding extra steps for users.
- Data Governance: Existing data masking, row access policies, and governance rules automatically apply to every agent interaction — so sensitive data stays protected without additional engineering effort.
- Verified Answers: Insights are grounded in real, query-based results that produce trusted outputs.Engineering teams can boost confidence by providing pre-verified queries for commonly asked questions. When the AI sees these trusted examples, it learns to apply the same logic to similar questions, improving accuracy and reliability across the board. This ensures that users can trust the AI while exploring data on their own.
This means any user can harness the power of AI while staying within the security perimeter of their Snowflake data platform.
Powered by Snowflake Cortex
Under the hood, Snowflake Intelligence leverages Cortex Agents — including Cortex Analyst, Cortex Search, and custom agentic tools — to orchestrate and execute SQL securely on behalf of users.
The platform also integrates Snowflake’s data_to_chart tool for rapid visualisation, and supports cross-region inference for optimal performance and model choice flexibility.
Models available
Choose from industry-leading models such as:
- Claude 4.0, Claude 3.7, Claude 3.5
- GPT-4.1
All are available directly inside Snowflake through Cortex, without leaving your secure environment. Full list available here
The Snowflake advantage
Snowflake Intelligence isn’t just another AI interface — it’s a unified, governed, and production-ready AI platform.
It shortens the path from question to insight to action, making AI adoption faster, safer, and more impactful for every organisation.
For companies in Australia and globally, this means less time building infrastructure and more time unlocking value – securely, at scale, and conversationally.
If you’re interested in trying out Snowflake intelligence this page is a great place to start or reach out to us via this link
Why Mantel
Mantel is an Elite Snowflake Partner and was the ANZ Partner of the Year 2025. We have a proven track record of building and scaling GenAI applications for clients. Implementing AI at scale requires more than technology – it demands a future-minded approach that balances people, processes, and platforms.
Mantel helps organisations rapidly prototype, experiment, and identify the most valuable AI use cases while ensuring cost and performance are managed effectively. We then deploy these to production in modular, scalable ways. We also have extensive experience establishing data platforms, building robust and well-governed data pipelines, creating semantic data models that reflect business logic and concepts, and generating insights from data.
By combining our AI and data expertise, we ensure that your data-driven AI solutions deliver lasting value and position your organisation for sustainable AI growth.
