Fnac Darty Streamlines Vendor Collaboration with AI
As a Partner and AI, Agentic & Data Strategy Specialist at Converteo, Guilhem Bodin advises Marketing and Digital executives on transforming their operations. He focuses on designing agent-driven strategies, optimizing customer journeys for the conversational era, and measuring the business impact of AI investments.
Key Takeaways:
- A single source of truth: The semantic layer standardizes the data language between the retailer and the vendor, eliminating technical friction.
- From visualization to recommendation: Integrating conversational AI slashes analytics time from hours to mere seconds.
- Security and quality control: Pairing strict access management with an “LLM-as-a-Judge” guarantees the reliability of the generated insights.
Historically, omnichannel brands face massive latency when trying to access their sales data. Operations often rely on static dashboards or Excel files that arrive weeks late. Meanwhile, e-commerce giants have established a new baseline: instant, actionable access to information. To level the playing field and unlock the full power of its omnichannel ecosystem, Fnac Darty completely overhauled its retail data sharing strategy. They redesigned their “My Retail Link” platform from the ground up in partnership with Converteo.
The Limits of Early Retail Data Sharing Models
Four years ago, Fnac Darty rolled out the first iteration of its vendor data-sharing platform. While the initiative addressed a clear need for transparency, day-to-day use quickly exposed operational friction points. Vendors and internal teams were running off different calculation baselines, effectively turning strategic meetings into data analyst debates over which numbers were right. Exposing massive volumes of data inevitably creates complexity. To actually trigger concrete commercial actions, the data must be instantly digestible.
The Tech Foundation for a Secure Ecosystem
Rebooting My Retail Link required strictly defining user journeys and KPIs. The entire infrastructure runs on Google Cloud: BigQuery centralizes the data, Dataform preps the tables, and Looker handles visualization across 11 business dashboards. The cornerstone of this architecture is the semantic layer (LookML), which provides a single, unambiguous definition for every metric.
This rigor establishes a true single source of truth and enables ultra-granular access governance. Every vendor sees only their specific operational scope, guaranteeing absolute confidentiality regarding competitor data.
Conversational AI Agents: The New Analytics Standard
Platform users have highly diverse profiles. A CMO lacks the time and technical bandwidth to cross-reference filters across multiple dashboards. To tear down this barrier, Fnac Darty integrated “Mira,” a conversational agent powered by Gemini. Users simply ask questions in plain English: “Why did my sales drop 10% in Paris yesterday?”
In seconds, Mira cross-references inventory, pricing, promotional pressure, and competitive context. The tool then instantly outputs a ready-to-use analysis and highlights the exact levers to pull.
Governance and Trust for Large-Scale Deployment
Integrating generative models with critical business data demands absolute reliability. To eliminate hallucination risks, the teams deployed an “LLM-as-a-Judge.”
This additional mechanism evaluates the relevance and mathematical accuracy of the agent’s answers before they ever reach the user. This security layer builds the trust necessary to drive business-side adoption. The platform already boasts over 400 active users and is targeting a massive rollout to 3,000 vendors. Finally, brand meetings can move past endless number-crunching and focus on product assortment strategy and ROI.
Watch the full keynote and see a live video demo of the tool by checking out the RUNTIME.AI replay on our YouTube channel.
