Summary
In this chapter, we discussed the integration pattern for building a real-time intent classification system using Google’s Gemini Pro generative AI model. We started by introducing the concept of real-time integration patterns, which prioritize low latency over efficiency and volume, as opposed to batch-processing integration patterns.
The use case we developed is an e-commerce company that wants to improve its customer service experience by automatically categorizing incoming customer inquiries into predefined intents, such as order status, product inquiry, return request, or general feedback. This classification can then be used to route the inquiry to the appropriate team or provide automated responses for common issues.
The architecture proposed is a serverless, event-driven architecture on Google Cloud, consisting of an ingestion layer (Cloud Functions), an AI processing layer (Vertex AI with Gemini Pro), an intent classification model, orchestration and...
