AI
For Peugeot, we built an AI chatbot made for car buyers: it answers fast, guides people to the right model, captures leads, and keeps simple questions away from the support team.
Key Takeaways
Takeaway Reality built a custom AI chatbot for Peugeot that guides car buyers to the right model, answers questions in seconds and captures leads on the Peugeot website.
The chatbot reached over 750,000 visitors and started more than 21,000 conversations, so roughly one in 35 visitors chose to talk to the brand.
Visitors exchanged 65,000+ messages, about three per conversation, which shows buyers use an automotive AI assistant for real follow-up questions.
Peugeot’s support team gets fewer simple tickets because the chatbot now handles repeat questions about models and trims.
The chatbot never quotes prices and sends buyers to a dedicated pricing site instead, which tripled that site’s users.
All data and chat logs are stored on TAR’s own EU-compliant infrastructure.
TAR’s AI chatbot development services have also delivered internal AI assistants for Fiat, Alfa Romeo, Jeep and Leapmotor.
Background
Peugeot's website gets heavy traffic, but most visitors look around and leave without ever talking to the brand. Each of them had a question. Nobody was there to answer it.
The people who did reach out went to customer support. Much of that was simple, repeat questions: which model fits a family, what the trims include, where to see prices. Support spent its time on those instead of the tickets that really needed a person.
On top of that, Peugeot has a large amount of product and brand material. Any chatbot had to find the right answer in all of it, quickly, without making the user wait.
Project
What we built: a custom AI chatbot for Peugeot
We did not plug in a generic bot. We built one around how people actually shop for a car.
Trained on Peugeot's own material. We gave the chatbot everything it needs to answer for the brand: models, features, trims and brand information.
Fast answers from a big database. The training data is large, but we found a way to pull the right answer from it directly. Users get a reply in seconds, not after a long pause.
Guides, never guesses. The chatbot does not quote prices, because prices change. Instead it helps the user narrow down what they need, then links them to Peugeot's site for live, up-to-date pricing.
Lead capture built in. While helping, the chatbot can collect contact details from interested buyers, so a quick question can turn into a sales conversation.
Simple for the user. Open it, ask in your own words, get an answer. No menus, no forms to fill first.
Security for the brand. All information, training materials and chat logs are securely stored on our own infrastructure, which complies with EU law.
Performance & statistics
Where visitors go
TAR chatbot analytics · all-time totals, rounded down
Before the chatbot, those 21,000+ visitors had no one to ask. Now each of them stayed, asked a question, and got pointed to a model, to live pricing, or into a lead form.
Support: fewer simple tickets, more time for real problems
The chatbot now answers the basic, repeat questions that used to land in Peugeot's support inbox. Support spends its time on high-priority tickets instead.
This matches what the industry sees. IBM notes that when chatbots take routine questions, ticket volume drops and agents can focus on complex cases. IBM also cites a Gartner prediction that AI will resolve 80% of common customer service issues without a human by 2029.
From question to purchase
The chatbot cuts the path from "I'm just looking" to "show me the price" down to one conversation.
Buyer journey · static website vs. chatbot
Static website | With the chatbot | |
First step | Scroll menus and model pages | Ask in your own words |
Finding the right model | Compare pages yourself | Chatbot narrows it down with you |
Simple questions | Search, give up, or email support | Answered in seconds |
Pricing | Find the configurator on your own | Direct link to live pricing |
Visitor who isn't ready | Leaves, brand never knows | Leaves contact details as a lead |
This matters because most car buyers don't arrive knowing what they want. Cox Automotive's 2025 Car Buyer Journey Study found 71% of US buyers started with an open mind, and buyers visit 4.6 websites on average. Only 25% use automaker websites at all, so every visitor who does arrive is worth keeping.
3x more visitors to the pricing site
Once the chatbot went live, users on Peugeot's new pricing site tripled.
Peugeot had an explicit requirement: the chatbot must not give prices. So we built a separate site where all prices are listed. Because the site was new, traffic was low at first. When the chatbot started guiding buyers there, the number of users grew 3x.
Illustration · example conversation, not a real transcript
Long story short, a chatbot can drive traffic wherever you need it: to articles, to your blog, or straight to the page where you sell.
Why personalized beats static
Buyers expect to be treated as individuals, and brands that do it earn more.
Finding | Source |
71% of consumers expect personalized interactions; 76% get frustrated when they don't get them | |
Personalization most often lifts revenue by 10–15% | |
76% say personalized communication was a key factor in considering a brand | |
Car buyers name real-time answers and personalized recommendations as the top benefits of AI tools | |
84% of mostly-digital car buyers who used AI assistants were highly satisfied with the purchase |
A static page shows everyone the same thing. The chatbot answers each visitor's own question, which is exactly what this research says buyers want.
Results
Over 21,000 visitors who would usually have left without a word started a conversation with Peugeot instead.
750,000+ website visitors had the chatbot available.
21,000+ conversations started. That is roughly 1 in every 35 visitors choosing to talk to the brand.
65,000+ messages exchanged, about 3 per conversation. People are not just clicking once; they ask follow-ups and get real help.
Fewer simple support tickets. Repeat basic questions are now answered by the chatbot, so the support team can focus on high-priority cases.
Every one of those conversations is a visitor who learned more about the brand, got pointed to the right model, or left their details as a lead.
How our AI chatbot development services work
Every step of this project was planned with Peugeot, around Peugeot's needs.
Understand the business. What customers ask, where they drop off, what support is overloaded with.
Set the rules. What the chatbot should answer, what it should hand off, and what it should never say (like prices).
Prepare the knowledge. Collect and structure the brand's material so answers are accurate and fast.
Build and test. Tune answers, speed and tone until it feels like talking to a well-informed Peugeot advisor.
Launch and improve. Watch real conversations and keep refining.
Proven across the automotive industry
Peugeot is not our only automotive client. We have built similar internal AI assistants for Fiat, Alfa Romeo, Jeep, Leapmotor and more, helping their teams find the right information faster and improving efficiency across the business.
Looking for AI chatbot development services?
If thousands of people visit your site every month, most of them are leaving with unanswered questions. Our AI chatbot development services give you a chatbot built for your products, one that turns some of those visitors into conversations, and some of those into customers, with a more personal experience for every visitor.
Curious how many conversations your site is missing? Let's talk.
How do I know which AI solution is right for my business?
Is an AI chatbot GDPR compliant, and where is the data stored?
How much does it cost to build an AI chatbot for a business?
Can a custom AI chatbot be trained on our company’s own data?
What are AI chatbot development services?
Can't find the question you are looking for?
Send us a message and we'll get back to you as soon as possible!
Tell us where your AI is stuck.
An idea, a stalled pilot, or a phone that won't stop ringing unanswered – the first call is a working session, not a pitch.
