Streaming Data for Real-Time AI Applications

Chapter 3. Real-Time AI in Industries

AI is already impacting businesses in dramatic ways. The use cases that follow all share a general pattern: they identify data and collect data for real-time AI, which involves capturing relevant inputs. Once the data is collected, the AI models and processing techniques are selected, ranging from machine learning algorithms and deep neural networks to rule-based systems and reinforcement learning, depending on the complexity of the analysis required. The AI system then generates outputs, which may take the form of predictive analytics, automated decisions, or adaptive responses that modify system behavior in real time. All of these use cases have real-world business impacts that show how AI transforms data into actionable intelligence, optimizing processes and driving innovation.

The use cases follow the same basic pattern outlined in the opening section of the report, with a producer, broker, and consumer. Chapter 1 explained how the customer service agent system follows this basic pattern, but in general it goes as follows:

1. The producer identifies and streams relevant data; AI may filter noise to focus on significant events.

2. The broker performs real-time analysis, transforming, enriching, and applying AI techniques (e.g., natural language processing [NLP], computer vision, predictive analytics) for deeper insights.

3. The consumer receives the results, which may be an AI system (e.g., chatbot, fraud detection) or a human operator; results ...