Conversational In-Сar AI Assistant for Next-Gen EVs

Project snapshot

The client is a global luxury EV brand developing next-generation electric vehicles, advanced battery technologies, and user-focused in-car experiences. Intellias helped the client build a scalable conversational AI-powered car assistant that brings LLM-powered interaction into the EV digital cockpit.

The eMobility solution includes a multi-layered navigation module using LLM APIs for intent detection, data extraction, and contextual feedback. It supports dynamic routing, POI discovery and recommendations, spatial awareness, and natural conversational responses, enabling drivers to interact with navigation and cockpit services beyond rigid voice commands.

Business challenge

The client needed to reduce reliance on costly third-party voice assistant and navigation vendors by building a scalable in-house solution. Beyond cost optimization, this meant gaining strategic ownership over the product roadmap, user experience, data flows, and future AI-enabled digital cockpit capabilities.

For a premium EV brand, the in-car AI assistant had to go beyond standard voice control. It needed to understand context, support real-world navigation scenarios, respond naturally, and integrate smoothly with vehicle systems while meeting automotive-grade standards for performance, reliability, and scalability.

The key challenge was to create a flexible architecture for immediate navigation use cases while keeping the AI-enabled mobility solution ready for future cockpit domains, including car commands, user manuals, calendar integration, and connected services.

Solution

Intellias set up a team that covered full product engineering lifecycle, including UX/UI design, solution architecture, backend and SDK development, integration, testing, continuous improvement, and scaling.

To accelerate development and bring AI-powered conversational capabilities into the digital cockpit, Intellias adopted ChatGPT as the core language model for the in-car AI voice assistant. This enabled the team to prioritize feature rollout, fast iteration, and production readiness instead of investing heavily in early-stage model fine-tuning before the product value was validated.

The final solution included a Kotlin-based SDK integrated with the Android-based IVI system and a cloud-hosted backend on Microsoft Azure. The architecture was designed to connect LLM capabilities with in-vehicle navigation logic, allowing the car AI assistant to process natural language requests, identify driver intent, extract relevant data, and provide smart feedback within the cockpit experience.

The product roadmap was built around modular expansion. Initial work focused on conversational navigation, route-related requests, POI search, recommendations, and spatial awareness. The architecture also created a foundation for turning the solution into a broader AI assistant for a car, covering future digital cockpit domains such as car commands, vehicle manuals, calendar services, and other AI-enabled in-car experiences.

Business outcomes

The solution brought natural AI-powered voice interaction into the client’s EVs, improving the quality of the driver experience and making navigation more conversational, adaptive, and user-friendly.

By moving from vendor-based solutions to an in-house AI-powered car assistant architecture, the client gained stronger control over product evolution, UX decisions, data management, and long-term feature development. This created a strategic foundation for continuous experimentation and faster deployment of new digital cockpit capabilities.

The scalable architecture allows the client to expand the AI voice assistant car integration across new use cases and vehicle services without rebuilding the core platform. It also supports a more flexible approach to innovation, enabling the client to test, refine, and release new AI-driven cockpit features faster.

The AI car assistant app also delivered a threefold decrease in operational costs compared with vendor-based alternatives, while giving the client greater ownership over its AI roadmap and in-vehicle digital experience.

Key impact:

• 3x decrease in operational costs compared with vendor-based solutions
• Scalable in-house architecture for future digital cockpit expansion
• Natural conversational navigation adapted to real-world driver requests
• Greater strategic ownership over data, design, UX, and product roadmap
• Faster experimentation and feature rollout through modular architecture
• Foundation for future AI-enabled cockpit services, including car commands, manuals, calendar integration, and connected vehicle experiences