The growth of Agentic BI is accelerating thanks to standards and open source technologies. With players like Dremio, dbt Labs, and Snowflake, companies are building data intelligence that is sovereign, interoperable, and transparent—a next-generation BI capable of harnessing the potential of AI-powered agents.

Today, artificial intelligence is transforming business uses of data: it no longer simply produces reports, but is becoming capable of understanding, interpreting, and making decisions. This is the principle of Agentic BI, where intelligent agents can interact with data, generate insights, and support decision-makers.

But this promise only makes sense if it is built on solid foundations: clear governance, open standards, and an interoperable ecosystem.

This is where open source plays a central role.

Recent initiatives, such as the open source of dbt Labs' MetricFlow semantic engine (Apache 2.0 license) and the Open Semantic Interchange (OSI) initiative led by Snowflake, reinforce this vision.

For its part, Dremio powers its Intelligent Lakehouse on open standards such as Apache Arrow, Apache Iceberg, and Apache Polaris, which help realize the promise of intelligent, truly agentic BI.

Agentic BI: towards more autonomous data intelligence

Agentic BI (or AI-powered Business Intelligence) marks a major breakthrough in the way companies interact with their data.

It goes beyond traditional business intelligence tools (like Power BI or Tableau), which still rely on manual dashboards or static reports.

Here, business users no longer need to code or manipulate complex interfaces. They can communicate in natural language (natural language interaction) with agents capable of understanding a business question, analyzing raw data, and automatically generating actionable insights.

For example, a sales manager can ask:

"What is our average revenue by region this quarter?"

And the agent responds immediately, drawing on real-time data from the company's data warehouse.

This new way of working makes analysis more fluid, reduces reliance on BI experts, and accelerates decision support.

But this apparent autonomy is only possible if business semantics, data governance, and technical consistency are rigorously aligned.

BI Agents must rely on a solid semantic layer, ensuring that each indicator (KPI) retains the same meaning, whether used by a dashboard, an API, or a chatbot.

AI is only valuable if the data it manipulates is reliable, traceable, and standardized.

It is precisely in this area that the open source ecosystem provides a sustainable and sovereign solution.

Open-source to standardize the semantic exchange layer and speak the same language between BI and AI

In an environment where a multitude of tools and platforms coexist, semantic standardization is becoming a key issue.

dbt Labs' MetricFlow engine, now open source under the Apache 2.0 license, embodies this advancement. It centralizes the definition of business indicators in a single semantic layer, ensuring perfect consistency throughout the data lifecycle.

Each metric, such as revenue, conversion rate, or number of customers, is defined once and then automatically translated into consistent and auditable SQL queries.

This eliminates discrepancies between BI tools and increases transparency for teams.

This approach is part of the Open Semantic Interchange (OSI) initiative, led by Snowflake and its partners.

The goal: to create an open standard that allows all systems—whether BI, AI, or Data Science—to speak the same language and ensure that each KPI follows the same logic, wherever it is calculated.

Thanks to this foundation, intelligent agents can integrate directly into the data value chain without compromising reliability or consistency.

This is an essential condition for building trust in the use of Agentic BI and predictive models based on real business data.

The Agentic Data Platform to support the data lifecycle

In this new data landscape, Dremio occupies a strategic position.

Its Intelligent Lakehouse platform is built on robust open source foundations, facilitating the connection, virtualization, and querying of all data sources, whether on-premises or in the cloud.

• Apache Arrow: an in-memory columnar format that significantly accelerates calculations and standardizes exchanges between analysis systems.
• Apache Iceberg: an open transactional format for data lakehouses, ensuring reliability, versioning, and data governance at scale.
• Apache Polaris: a unified and open metadata catalog, facilitating the discovery and traceability of datasets.

These components make Dremio much more than a query engine: it's a modern Open Data Platform, capable of connecting data without moving it, while preserving sovereignty and performance.

It also offers a native semantic layer that integrates with AI agents' MCP connectors to ensure seamless and controlled data exploitation throughout the ecosystem.

This approach heralds a major evolution: the birth of the Agentic Data Platform, an infrastructure designed to support the AI-powered and real-time uses of tomorrow, while maintaining the governance and transparency of open source models.

 

An open ecosystem that showcases Smile’s integration expertise

Open-source and standards aren't just technical choices: they're strategic ones.

In a world where systems must understand and cooperate, open source is becoming a pillar of trust, resilience, and transparency.

The integration of these building blocks into a single ecosystem opens up new perspectives for organizations looking to industrialize Agentic BI while maintaining control over their data.

• dbt / MetricFlow provides semantic clarity: metrics are documented, shared, and interpreted uniformly.
• Dremio, thanks to Arrow, Iceberg, and Polaris, offers flexibility and sovereignty: data remains in open formats, queryable via the MCP, for seamless integration with intelligent agents.

This allows a company to build hybrid data intelligence, where BI agents interact directly with the semantic layer and distributed data sources.

Analysis no longer depends on proprietary models or manual reporting processes: data becomes alive, traceable, and usable in real time.

This new architecture enables explainable and auditable AI, capable of helping decision-makers better understand key indicators and generate insights relevant to the real world.

By leveraging open and interoperable technologies, companies reduce their dependence on vendors while developing new synergies between their BI, Data, and AI teams.

This is a concrete way to combine innovation, trust, and digital autonomy.

Agentic BI is therefore not limited to more efficient conversational agents. It embodies a profound transformation in the way organizations understand and leverage their data.

A sovereign, explainable, and open BI, aligned with Smile's values:

• Open rather than silo,
• Standardize rather than lock down,
• Innovate with complete transparency.

Contact our Smile experts to explore how to build your own Agentic Data Platform and realize the potential of Agentic BI to serve your business.
 

Glossary of Agentic BI and Agentic AI

• Agentic BI: Business intelligence augmented by autonomous agents capable of querying, analyzing, and providing actionable insights.
• Agentic AI: Intelligent agents using AI to understand and interpret data, generate decisions, and automate analyses.
• AI-powered BI: Analyses and dashboards assisted by artificial intelligence, enabling the automatic generation of insights.
• Raw data: Unprocessed data, collected directly from sources, before any structuring or processing.
• Traditional business intelligence: Classic BI methods based on static reports and manual reporting.
• Natural language interaction: The ability to interact with data systems in natural language, without technical queries.
• Real-time data: Information accessible and actionable immediately after creation or collection.
• Data warehouse: Centralized infrastructure for storing and organizing data for analysis and reporting.
• Decision support: Systems or tools that facilitate decision-making by providing reliable and relevant analyses.
• Predictive models: Analytical models capable of anticipating future trends or events based on historical data.
• Generate insights: Automatic extraction of analyses, indicators, and actionable recommendations for the business.
• Decision makers: Individuals or teams responsible for making strategic decisions based on the data and analyses provided.
• Traditional BI tools: Traditional reporting and analysis software, such as Power BI or Tableau, often dependent on manual intervention.