AI vs. Karen - Real-Time Customer Sentiment Detection at Scale

Explore building AI-powered real-time streaming pipelines for customer sentiment detection in this conference talk from NDC Copenhagen. Learn how to architect systems that can identify difficult customers before problems escalate by implementing real-time sentiment analysis on live unstructured user data from your own systems or social media platforms. Discover the essential architecture, libraries, and tools needed to perform sentiment analysis using Apache Kafka and Apache Flink, while understanding the challenges of integrating AI models into streaming technologies. Walk through a practical, real-world example that demonstrates the implementation process from start to finish. Compare the differences between powerful large language models (LLMs) and lightweight small language models (SLMs), examining the pros and cons of each approach to help you select the best solution for your specific use case. Gain the foundation, insights, and inspiration needed to implement real-time sentiment analysis systems whether you're experienced in streaming systems and data engineering or simply curious about applying AI to solve real-world customer service challenges.