the case of E.Leclerc SAV Ouest

In a context of rapid growth and increasing pressure on logistics costs, the entity E.Leclerc SAV West launched the project “Towards artificial intelligence”with the support of You and Digital’Ease.

Objective: evaluate, test and implement concrete uses of artificial intelligencecapable of improving the productivity of the after-sales service, in compliance with company constraints, organizational culture and the safety requirements imposed by GALEC.

The results are significant: a validated use case with high logistical impact, an increase in maturity on AI productivity tools (Gemini, NotebookLM), and a clear vision of the solutions to be industrialized in the months to come.

Context and strategic issues

1.1. E.Leclerc SAV Ouest: a rapidly growing service

The entity manages the after-sales service for the E.Leclerc stores in the Grand Ouest, which cover three purchasing centers:

Its economic model is based on a system of voluntary contributions member stores.

In three and a half years the organization went from From 41 to 91 storeswith a projection of From +20 to +30 subscriptions in the coming years.

It is based on:

• a central place,
• two satellite product collection sites,
• 20 technicians ensuring interventions.

In addition to after-sales service, the entity also manages:

• fuel oil distribution,
• a new product delivery/installation business, currently unprofitable.

1.2. Greater logistical pressure

THE transport and the computer processing represent the highest expenditure item.

As the volume of articles covered increases, the challenge is clear:

reduce unnecessary transport, speed up processing times and make flows more reliable.

Various actions have already been undertaken: modernization of the premises, reorganization of flows, generalization of the showers for the traceability of goods.

The project “Towards artificial intelligence” he therefore aimed to take a further step forward.

1.3. Objectives of the AI ​​project

The ambition was twofold:

1. Identify useful, profitable, and immediately achievable AI use casesbeyond simply writing emails with Gemini.
2. Respect security and governance constraints data from GALEC and avoid any overlap with the group’s national experiments.
3. Analysis of experiments: successes, limitations and lessons of artificial intelligence

2.1. Use case no. Tip #1: Instantly Identify “Low Value” Products Using “Vibe Coding”

The business problem

Some categories of products received in store are automatically considered as not repairable and give rise to a credit.

Today they are still sent to the central site, generating:

• of unnecessary transportation,
• AND significant logistics cost,
• AND additional processing time.

First attempts: conversational AI

Querying models like Gemini or Claude with Excel files showed:

• operational slowness,
• file size limitations,
• insufficient precision.

The break: «Vibe Coding»

The use of Co-pilot GitHub with Visual Studio Codeusing an AI-assisted development approach, was decisive.

By simply describing the desired application, the AI ​​generated:

• architecture,
• the code,
• the web interface,
• database management.

Result

A Functional internal web application was created in a few hours :

• by entering a barcode,
• by querying an internal database,
• immediate response “low value / not low value”.

The internal test was described as “fully satisfactory”.

THE distribution is planned from a hundred shops.

Expected impact: Massive reduction in unnecessary transportation.

2.2. Use case no. 2: Route Optimization: Humans are still ahead, but AI is making progress

The necessity

E.Leclerc SAV Ouest must organize different types of tours:

• shuttles to shops,
• fuel oil deliveries,
• installations of new products,
• technician interventions.

What artificial intelligence can do today

AI has shown real value for:

• automatically extract information contained in the delivery notes in PDF format,
• standardize addresses,
• facility planning data.

What AI still can’t do as well as humans

For thefine optimization of a daily tourhuman skills remain superior, in particular with regard to:

• the integration of the territory,
• management of implicit constraints,
• dynamic priority.

Perspectives

1. The new AI module of the enterprise software (S9000) will be tested at launch.
2. The partnership with Leclerc Woos folder paves the way for a tailor-made instrument:
   • automatic calculation of distances,
   • invoicing of out-of-scope deliveries,
   • future integration into a business tool.

2.3. Use case no. 3: Office AI and knowledge management – ​​an immediate productivity lever

Gemini (Google Workspace): Huge time saver

Gemini is allowed to:

• synthesize complex sets of emails,
• prepare complete files,
• write professional letters (including payment reminders),
• standardize the tone and quality of communications.

NotebookLM: the reveal of the project

NotebookLM, based on a closed corpus of documents, provided:

• AND precise question with citations to the sources,
• A structured analysis agreements with suppliers,
• the automatic generation of:
  • presentations,
  • explanatory videos,
  • podcasts,
  • teaching materials.

Concrete applications to be industrialized

• creation of a knowledge base for agreements with suppliers,
• continuous training for point of sale contacts,
• production of training content for after-sales service teams.

2.4. Use case no. 4: Searching for technical information: a current limitation of artificial intelligence

Test via Perplexity, Gemini and ChatGPT to obtain:

• exploded plans,
• spare parts references,
• multi-brand information

they were without successmainly due to:

• lack of access to password-protected sites,
• anti-robot protection,
• the scarcity of reliable data in open data.

Conclusion: no added value of AI for this use case in the short term.

Technologies and methodologies analyzed

Summary table of the tools tested and their uses:

Action plan: industrialization and ramp-up

4.1. Operational deployment (priority ★★★)

• Finalize the after-sales service sorting application.
• Distribute the tool in stores across the network.
• Industrialize EAN file production for maximum reliability.

4.2. Logistics optimization

• Try the new AI module of S9000 business software (★★★).
• Automate the reading and integration of PDF delivery notes (★★★).
• Consider a distance calculator app based on Woosmap (★★☆).

4.3. Knowledge management and productivity

• Integrate Gemini and NotebookLM into everyday uses (★★★).
• Build a documentary base of agreements with suppliers (★★★).
• Produce interactive training modules for store teams (★★★).

Conclusion: Artificial intelligence as a performance driver for operational organizations

The project “Towards artificial intelligence” demonstrates that a pragmatic approach, centered on business needs, allows you to:

• generate quick wins,
• automate low-value tasks,
• implement operational tools in a few hours,
• knowledge of the structure,
• respecting safety requirements and human realities in the field.

With this first cycle of experiments, E.Leclerc SAV Ouest positions itself as a pioneer actor in the rational use of artificial intelligence in the after-sales services and logistics sector.

And this is just the beginning.

Giving meaning to digital transformation

Artificial intelligence does not replace humans.

She him it gives back time, clarity and freedom.

Chez You and Digital’Easewith the program Focus on artificial intelligencewe help leaders transform their business respecting their pace, their values ​​and their teams.

Do you want to start your digital transformation with trust and humanity?

We discuss your needs and explore first lines of action.