-
-
Enter your booking ID to instantly pull up your rental.
-
Choose your upgrade journey: AI-guided or manual selection.
-
Meet the Sixt AI Sales Agent: ready when you are.
-
Hands-free conversation: the agent is now listening to your needs.
-
Personalized recommendation in seconds: your ideal PEUGEOT 408 upgrade.
-
Booking confirmed. Digital key and full details in one clean view.
-
Vehicle unlocked and ready to drive with a single tap.
-
Print-ready confirmation for customers who still love paper receipts.
Inspiration
Sixt has incredible cars and upgrade options available at every branch, but upselling depends heavily on staff availability, language skills, and time. We wanted to solve two problems at once: give every customer a personalized, human-like upgrade experience, and ensure branches can scale that experience globally without additional workload. Seeing how generative AI can adapt to customer needs in real time motivated us to build a digital sales agent that feels natural, understands context, and drives meaningful revenue - anytime and anywhere.
What it does
Our Digital Sales Agent turns the entire upgrade process into a fully automated conversation. Customers simply enter their booking number, and the AI agent interacts with them just like a real Sixt employee, asking about passengers, luggage and preferences. It then recommends the best-fitting vehicle, offers personalized upgrades or protections, and ultimately enables customers to unlock their car through the app with a single tap. It’s available in 50+ languages, works across branches worldwide, and ensures that every booking receives the perfect offer.
How we built it
We built our Digital Sales Agent as a cross-platform solution available both on the web and on iOS. On the web, we developed a React frontend connected to a Python backend that handles all logic, API routing, and conversation processing. In parallel, we created a fully native iOS app using Swift to deliver a fast, seamless mobile experience that integrates directly with the Sixt-style interface. For the intelligence layer, we connected the system to Gemini, which powers the conversational agent and customer-profiling logic. To make the interaction feel human and engaging, we used ElevenLabs for high-quality speech output.
Challenges we ran into
One of the biggest challenges was shaping the AI’s behavior so it followed our intended user journey. Getting the voice agent to ask the right questions, maintain context, and upsell naturally, without sounding repetitive or pushing irrelevant options, required a lot of fine-tuning. Another challenge was making the AI consistently recommend the right car. We had to engineer prompts, constraints, and backend logic so the model only offers vehicles that are actually available and aligns them with the customer’s needs (passengers, luggage, comfort preferences, etc.).
Accomplishments that we're proud of
We’re proud that we managed to build a fully working Digital Sales Agent across two platforms, a web app and a native iOS app, within the hackathon timeframe. Both versions follow the same user journey, and deliver a polished end-to-end experience. Building a system that works in 50+ languages and can scale globally is something we’re especially excited about.
What we learned
We gained valuable experience integrating LLM models, voice technology, and backend logic into a smooth multi-platform user journey. Throughout the project, we learned how powerful, but also how sensitive AI-driven customer interactions can be.
What's next for CamelCase Crew
We want to improve our agent's customer understanding so that the agent can learn from customer preferences, booking history, and behavior patterns to make even smarter and more contextual upsell suggestions. We also aim to enhance the experience with richer generative UI elements that would dynamically generate car comparisons, personalized offers, next-best recommendations, and visual explanations that adapt to each customer’s profile.
Log in or sign up for Devpost to join the conversation.