The Carrefour Group currently has more than 50 AI applications in production, across areas including assortment, pricing, promotions, the supply chain and customer recommendations.
Today, the Group is taking a new step in its AI transformation by deploying Copilot Expansion, a conversational AI agent capable of accelerating store opening decisions. By reducing the timeframe required to obtain a market study, this project illustrates the shift from analytical AI to AI integrated into business processes.
Challenge: Accelerating store opening decisions
Carrefour opens around 100 stores per year. Each new location brings together three key stakeholders:
- The Expansion team, which identifies opportunities on the ground;
- The Geomarketing teams, which produce detailed analyses to assess commercial potential;
- The Finance teams, which validate the investment. Administrative back-and-forth exchanges complete the process.
Consequence: The decision of whether or not to establish a new store could take up to a few weeks to collect data, calculate ROI, and evaluate the suitability of the site. In an ultra-competitive market, such a delay can lead to missed opportunities and economic losses.
“Agentic AI opens a new phase: it is no longer just about analyzing data, but about transforming the way teams work and make decisions.”
Edouard de Mézerac, Group CEO, Artefact
Solution: Copilot Expansion, an AI agent integrated into Carrefour’s specific business processes
Artefact and Carrefour have developed an AI agent capable of producing a complete market study in two minutes to evaluate the suitability of a store opening.
The agent analyzes the project zone and its competition, the socio-demographic profiles of the population, the revenue potential by neighborhood, and verifies the project’s financial viability by examining CAPEX, OPEX costs, and projected revenues.
The user experience is simple. An expansion manager places a point on a Google Maps-type interface and provides details about the project: expected surface area, parking spaces, and qualitative information such as a new subway line or a real estate project. The AI agent interacts with the user, requests any necessary clarifications, then generates a market analysis and an assessment of the suitability of the new site in a single click.
Artefact supported Carrefour in defining the use case, redesigning the business process, and making architectural choices.
“The challenge was not to create just another agent, but to insert it into a real and measurable process. We prioritized a pragmatic approach: starting with the tools already in place and building an agent capable of integrating naturally into business workflows.”
Vincent Blaclard, Managing Partner & Retail Lead France, Artefact
Results: Measurable returns from the very first deployments
Initial rollouts of Copilot Expansion demonstrate concrete impacts on operational performance and decision-making:
- Market study time reduced from several months to two minutes for a complete analysis
- Significant improvement in the accuracy of the revenue prediction model, with a 15-point gain
- Up to 50% of dossiers filtered upstream thanks to the self-service model, allowing geomarketing and finance teams to focus on high-stakes strategic projects
Carrefour now experiences better allocation of internal resources and faster, more homogeneous decision-making across all store opening projects.
Method: Reinventing roles to create value
The approach adopted by the Carrefour and Artefact teams combines the direct integration of the agent into the interfaces already used by the teams and leverages the Google Cloud Platform environment to facilitate scaling. Above all, this project involved rethinking the way teams work on a daily basis.
Four lessons emerge from this real-world experience for the effective design and deployment of AI agents:
- Rework processes, not just the tool: The focus should be on reinventing work before turning to technology.
- Anticipate: Integrate agentic AI right from the design phase of the transformation, rather than adding it as an afterthought.
- Take a pragmatic approach to technology: Rely on the existing ecosystem to facilitate adoption and reduce friction points.
- Don’t fall into the “all-agent” trap: Each need requires the right lever, whether it be traditional AI, linear processes, business intelligence, or autonomous agents.
“AI doesn’t do everything end-to-end. People always intervene at the end of the process, with a focus on empowering teams. The ‘human-in-the-loop’ approach remains essential to ensuring the reliability of results.”
Emmanuel Grenier, Executive Director of Supply Chain, E-Commerce, Data, and Digital Transformation
This case bears witness to the challenge of the coming years in the retail sector: sustainably anchoring AI at the core of operational decisions. The method developed by Carrefour and Artefact paves the way for a more agile organization, capable of accelerating its decisions while reinforcing human expertise.
Agentic AI thus becomes a major lever of competitive advantage for retail. The companies that succeed will be those capable of aligning process transformation, a robust data foundation, and the internal industrialization of AI agents.
During the Tech For Retail trade show on November 24 and 25, 2025, Carrefour and Artefact shared their feedback on the deployment of AI agents in retail professions. Access the replays (subtitles available in English):
Generative AI and Agents: The Next Revolution in Retail?
Speakers:
Emmanuel Grenier – Executive Director of Supply Chain,
E-Commerce, Data, and Digital Transformation
Edouard De Mézerac – CEO Groupe, Artefact
Pascal Clouzard — Co-founder of Tech for Retail and
former CEO of Carrefour France and Spain
How Can AI and Agents Help Speed Up Store Openings?
Speakers:
Arnaud Grojean – Chief Data & Analytics Officer, Carrefour
Vincent Blaclard – Managing Partner & Retail Lead France, Artefact
