Snowflake Cowork: The AI agent for querying your data
Snowflake Cowork transforms the way teams access their data. This video demonstration shows it in practice. Clément Lescure, Lead Analytics Engineer at MeltOne Advisory, presents the use case “Talk to my Data“. The objective is simple: query your data in natural language, without necessarily going through a traditional BI project.
Discover the demonstration by Clément Lescure, D&A Manager at MeltOne
Why decision-making remains slow
Today, most decisions still rely on traditional dashboarding. When business users look for the KPIs on which to base their analyses, they face three possible scenarios:
- First, when the question was anticipated during the dashboard scoping phase, the answer arrives quickly. The analysis is completed in 5 minutes. Indeed, one simply needs to consult the right KPI.
- Next, when the question/KPI was not anticipated. The data sometimes already exists in the dataset. However, a recalculation is often still necessary. As a result, the analysis takes an entire day, or even longer.
- Finally, the most complex case occurs when a complete study must be redone on an unforeseen subject. Two options are then available: either manually cross-reference data from emails, SharePoint, or PDFs, or restart a requirements scoping phase followed by a BI project. In both cases, the lead time rises to a week, sometimes much more.
In short: the further the question deviates from what was anticipated, the longer the response time. This is precisely what Talk to My Data seeks to reduce.
Snowflake Cowork: Snowflake’s new AI Agent
Snowflake Cowork, formerly Snowflake Intelligence, is the agentic component of the Snowflake platform. It allows users to interact with their data using natural language. The AI agent queries Snowflake warehouses and provides a contextualized response. No SQL queries are required. For data and BI teams, this changes the way unforeseen business requests are handled. For business users, it changes how they interact with data.
The demonstration scenario
The demonstration is based on a concrete financial use case. For example, the application combines accounting data, internal invoices, and external data from web services.
The presentation is structured into six parts:
- Presentation of the Snowflake Cowork AI agent.
- Ensuring response quality, a central challenge of Talk to my Data.
- Use of external data.
- Embedded application.
- From analysis to action.
- Security and FinOps.
Consequently, the demo covers the entire cycle. It starts from the question asked to the agent and goes all the way to the triggered action. In between, it addresses the reliability of the response.
Book an appointment
Let’s talk about your project and discover how MeltOne can turn your challenges into concrete, high-performing solutions.
