The End of Data Queues: How Alan Scaled Analytics with Dust
Alan, a €4 billion digital health insurer across four countries, needed to solve a critical challenge: enabling hundreds of employees to access data insights without overwhelming their data specialists. With operations spanning insurance, healthcare, and digital platforms, teams constantly required data for decisions—but relying on data experts for every query was unsustainable. We spoke with Alix Vermeulen (data team) and Maxime Faidherbe (operations) about how Dust AI agents are democratizing data access while maintaining quality and efficiency.
About Alan
Founded in 2016, Alan is a leading digital health insurance company operating across France, Spain, Belgium, and Canada. The company combines insurance services with a comprehensive digital health platform, connecting users directly to healthcare providers through intuitive interfaces. Alan's mission is to make quality healthcare more accessible and understandable for everyone. Valued at €4 billion, Alan serves thousands of companies and individuals with innovative healthcare solutions.
Challenge: Democratizing data access while maintaining quality
As Alan expanded across Europe, their data team faced mounting pressure to support business decisions across multiple departments. The team needed to solve two critical challenges simultaneously:
- Enable non-technical teams to access data insights independently
- Help data experts work more efficiently
The situation was complicated by:
- A complex database structure spanning 15+ schemas and hundreds of tables
- Team members with widely varying SQL expertise
- Documentation scattered across multiple platforms
- Senior data team members spending excessive time on basic SQL support
To better understand how this solution works across different user types, we spoke with two Alan team members who represent the primary personas benefiting from their Dust AI agents.
Alix Vermeulen works in Alan's data team, focusing on shopping analytics for glasses, contact lenses, and wellness products. As the sole data specialist in her cross-functional team, she uses the agent to accelerate SQL development and enable team self-service.
Maxime Faidherbe works in Alan's Operations team, focusing on insurance topics like affiliation and coverage. While he has solid SQL skills and strong data intuition, he faces typical business user constraints: limited time for complex queries and navigating multiple databases. He uses the agent to gain independent data insights without extensive time investment.
Solution: An intelligent SQL agent powered by Dust
Alan developed @Metabase, a sophisticated Dust AI agent that serves as an intelligent SQL specialist, helping team members navigate their Snowflake database and Metabase visualization tool. What makes this solution particularly powerful is its comprehensive integration with Alan's data ecosystem, thanks to native connections and the Dust APIs.
Powerful Knowledge Integration
The agent connects seamlessly with multiple data sources:
- Metabase dashboards and queries repository
- GitHub repositories containing table definitions and data models
- YAML files with detailed column descriptions and business logic
- Notion documentation for business metrics
How the Agent Works
The agent operates in three complementary modes:
- Dashboard Discovery First checks if existing dashboards answer the need - saving time and avoiding duplicate work. The agent quickly connects users to relevant pre-built analytics with proper filters and parameters.
- Query Generation For new analytics needs, the agent generates optimized SQL right in Metabase by understanding both business context and technical requirements. It follows Alan's best practices and includes clear documentation.
- Interactive Refinement Engages in dialogue to clarify requirements, suggest improvements, and troubleshoot results - much like working with an experienced data analyst. The agent is often fueled with the errors it gets from Metabase and it fixes them.
Results: Transforming data operations across Alan
Alan's implementation of Dust AI agents has transformed how their entire organization interact with data, delivering measurable improvements across three key areas:
Faster query generation
AI agents now handle the heavy lifting of SQL creation, dramatically reducing development time.
- Built-in understanding of table relationships and business logic
- Contextual assistance for complex data models
- Query development time cut from hours to minutes
"It's super helpful to get my first SQL - it has context on the data model. When I'm working on new tables I don't know at all, I ask the agent to create the query first, then I go and debug if needed." Alix, Data Team
Reduced data support burden
The data team can now focus on strategic analysis instead of fielding basic requests.
- Significant drop in data on-call requests, leading the data team to focus on complex analysis
- Faster response times for all data requests, helping non-technical teams leverage insights instantly
"Since I'm the only data person in my team, I encourage others to use the Metabase agent. It helps them become independent without having to ask me basic questions, and we've seen a clear reduction in simple SQL requests to the data team." Alix, Data Team
DIY data investigation
Teams can now explore unfamiliar datasets independently, unlocking previously inaccessible insights.
- Ability to work with unfamiliar tables without data team support
- More complex analyses now possible due to reduced friction
"Before, you would stick to simple queries because the cost of editing was too high. Now I can manipulate tables I don't even need to know about. What would have taken two hours before can be done in minutes - or wouldn't have been attempted at all due to complexity." Maxime, Operations Team
Widespread adoption
The launch of the @metabase agent has made the entire Alan team much more data-driven. Below are some adoption stats for the Metabase agent, and the numbers are climbing:
- 60% weekly active users in Operations team (up from 25% 1 year ago)
- 1/3rd of the entire Alan workforce, across all functions (Operations, Product, Data HR, etc.) uses it
Conclusion
The integration of Dust AI agents into Alan's data operations has eliminated common bottlenecks in SQL support while serving as a reliable tool for building queries. While the system doesn't achieve perfection, it has become an key resource - allowing the data team and experts to focus on key queries and monitoring while the AI handles routine data operations.
"It's mind-blowing. These are the kinds of use cases that are truly incredible and show the power of the Dust platform as an assistant builder.” Maxime, Operations Team