“We started with a shared email inbox where we managed customer requests, it was extremely time-consuming.”
10'000 tickets per month. 20'000 during peak activity. 7 languages. A shared email inbox as the starting point. That was the backdrop when Vinatis, one of Europe’s leading online wine retailers, began overhauling its customer service with fAibrik in 2020. Five years later, in May 2026, the company rolled out its third module: a conversational agent. Behind this decision lay a solid strategic vision, rigorous implementation, and above all, tangible results. Céline Boissin, Customer Service Manager at Vinatis, shared her experience during a webinar. This article gives you a sneak peek. To dive deeper and hear her full story, request the webinar replay below. Access the replay
Vinatis is one of Europe’s leading online retailers of wines and spirits. For over 15 years, the company has offered a selection of thousands of products to a passionate customer base, both individuals and professionals, across Europe. Its model is built on a carefully crafted shopping experience, from advice to delivery, with customer service at the heart of this promise.
Before implementing fAibrik’s solutions, Vinatis managed its customer service through a simple shared email inbox. This setup was reaching its limits. The Covid pandemic accelerated the crisis. Demand volumes skyrocketed within weeks, and agents were handling an average of 10'000 tickets per month. During peak periods, this number could double to 20'000 requests.
“We started with a shared email inbox where we managed customer requests, it was extremely time-consuming.”
This volume came with an added challenge. With nearly 20 countries served, the customer service team had to manage over 7 different languages. Manual handling was no longer sustainable. It was in this context that Céline Boissin began searching for solutions tailored to these challenges and turned to fAibrik.
After becoming a fAibrik customer in 2020, Vinatis didn’t deploy everything at once. The approach was gradual: identify needs over time, then add tools step by step. This phased logic allowed each solution to integrate smoothly into existing processes without disrupting the team.
The first module implemented was omnichannel ticketing. Emails, instant messaging, contact forms: all customer requests are now centralized on a single interface. Artificial intelligence analyzes every incoming ticket: it identifies the reason for contact, detects customer sentiment, assesses priority, and assigns it to the right advisor. The advisor has the full history at their fingertips to personalize their response. They can also rely on AI as a writing and translation assistant. Here, artificial intelligence acts as a personal assistant, never replacing human advisors. Thanks to this system, Vinatis was able to handle all requests, even during peak periods, with an average response time of under 24 hours.
Once demand management was stabilized, Vinatis wanted to better control its online reputation. Review management serves a dual purpose: acquiring new customers and retaining existing ones. By responding to negative reviews, the company resolves issues and recovers disappointed customers. By responding to positive reviews, it strengthens its relationship with its existing customer base. Vinatis now responds to all reviews received. Positive reviews are handled automatically. Negative reviews, which require more expertise and empathy, are managed manually by advisors. This blend of automation and human intervention is at the core of fAibrik’s approach.
The conversational agent was deployed in May 2026 as the third module of the collaboration with fAibrik. This conversational agent answers visitors’ questions in real time and natural language, 24/7. The goal is clear: relieve advisors of frequent, repetitive questions. This frees up their time to focus on more complex requests, those that require real human expertise. The AI agent relies on a knowledge base built and maintained by Vinatis’ teams. It doesn’t search the internet for information: it provides answers based solely on verified, internally validated data. This approach ensures responses are reliable and aligned with the company’s policies. Watch the replay
Rolling out a conversational agent that delivers real results doesn’t happen by chance. Here’s how Vinatis structured the project, step by step.
The first step to a successful deployment is scoping. While customer service is the natural lead for the project, other teams play a key role. The legal team can define what information the agent can, or cannot, share. The IT team can identify potential roadblocks with existing tools. Involving the right stakeholders from the start prevents surprises down the line. It’s also essential to clearly define the assistant’s scope: what questions should it handle, and what answers should it provide? For Vinatis, the initial priority was covering recurring questions. The agent will later expand to include delivery information. Finally, setting objectives and KPIs from the outset ensures the entire team can steer the project with clarity and take action to achieve the desired results.
The knowledge base is the foundation of the conversational agent. It’s the data the AI agent uses to formulate its responses. It was also the most time-consuming step for Vinatis, and that’s completely normal. The challenge is to gather all the information customers need, present it clearly, and structure it so it’s easily found and understood by the AI.
“Building the knowledge base took us a month, dedicating a little time to it every day.”
The knowledge base isn’t a static document. It’s a living resource that requires regular updates and enrichment. Questions asked by customers after deployment help identify missing information and refine the clarity of existing data. The more complete and well-organized it is, the more relevant the responses generated.
Before going live, several teams at Vinatis tested the agent by asking real questions. This phase is critical for two reasons. First, it verifies that the conversational agent responds correctly. Second, it highlights gaps in the knowledge base, so they can be fixed before customers encounter them. Only once testing is validated can deployment begin.
Going live marks the start of an active observation phase. The conversational chatbot answers visitors’ questions, and every interaction becomes a learning opportunity. The knowledge base can be adjusted to improve the accuracy and relevance of responses over time. The team can also identify unexpected topics visitors ask about and enrich the data accordingly. Deployment isn’t a finish line, it’s the starting point for continuous improvement, driven by data and customer feedback. That’s exactly what Vinatis’ experience illustrates.
Céline Boissin quickly saw the first effects of the conversational agent. Within weeks, the volume of repetitive questions handled by advisors dropped significantly. So much so that Céline redeployed some of the time saved to higher-value tasks.
“Advisors can now fully focus on serving customers.”
But the impact goes further. By analyzing questions asked via the AI agent, Vinatis identified areas of confusion on its own website. This feedback was shared with the relevant teams to improve content clarity. This virtuous cycle reduces questions at the source rather than just addressing them downstream. To discover all the results observed by Céline Boissin and learn how she manages the chatbot on a daily basis, request the webinar replay.
Deploying the conversational agent at Vinatis is a success because it delivers real benefits for both advisors and customers. But this article only tells part of the story. To get the webinar replay and hear Céline Boissin’s full experience, including the choices made at each stage, the challenges faced, and the lessons learned, click the button below. If you’d also like the complete Vinatis customer case study with all the data, request it below. Get the replay
What is Vinatis? Vinatis is one of Europe’s leading online retailers of wines and spirits. The company offers thousands of products to individuals and professionals across Europe, with a multilingual customer service at the core of its model.
What is a conversational agent? A conversational agent is an AI-powered tool that uses a company’s knowledge base to answer visitors’ questions on a website. In customer service, it handles frequent queries.
How does a conversational agent work? A conversational agent can operate in several ways: using a closed knowledge base, a decision tree, integration with customer service tools, or even internet searches. fAibrik’s solution relies on an internal knowledge base to ensure reliable, controlled responses.
What are the different types of chatbots? There are AI-powered chatbots, voice-based chatbots like callbots, FAQ chatbots that use pre-written responses, and non-automated assistants managed by humans. Each type serves different needs and contexts.
