Winning solution — Porsche Digital Campus Challenge (Customer Relations)

Porsche Digital — AI-Powered Customer Relations Hub

An AI-powered Customer Relations Hub that delivers real-time, personalized support to Porsche customers while enabling CR teams with faster resolution, full context awareness, and seamless human handover.

PythonAWSAI AgentVector DatabaseEmbeddingsRAGOmni-channel UX

Problem

Porsche customers expect fast, personalized, luxury-grade service, yet support experiences can suffer from fragmented knowledge, slower response times, and limited personalization. At the same time, Customer Relations teams need a centralized view of sessions and context to resolve issues efficiently without customers repeating themselves.

Solution

We designed an AI-powered Customer Relations Hub combining a customer-facing assistant with a CRH dashboard. The AI agent supports multi-modal input (text/image/voice), retrieves relevant context via semantic search, produces structured guidance, and escalates to human support when needed—preserving full context for a seamless handover.

Highlights

  • Winner — Porsche Digital Campus Challenge (Customer Relations category)
  • Designed an AI agent with semantic retrieval over Porsche knowledge sources
  • Multi-modal input: text, image, and voice
  • Human-in-the-loop escalation with full context preservation
  • CRH dashboard with KPIs, session tracking, and analytics

What this shows

Product thinking, structured execution, and practical engineering choices—especially around scalable backends, data workflows, and AI-enabled user experiences.

UI & System Diagrams

Selected visuals from the prototype and system design.

Customer-facing Porsche AI assistant UI

Customer-facing AI assistant integrated into the My Porsche app concept, featuring guided troubleshooting and human escalation.

Backend system architecture diagram

Backend and system architecture integrating AI agent, vector database, tools, and Porsche data sources.

AI agent data flow diagram

AI agent workflow: multi-modal input → semantic retrieval → response generation → satisfaction check → human handover.