博客來Engineering in Flow is a practical field guide for building production-ready generative AI systems in the era of agents, large language models, and human-in-the-loop design.
As GenAI tools collapse the distance between idea and execution, traditional software engineering habits are breaking down. Speed is no longer the bottleneck-reliability, control, and correctness are. This book shows experienced engineers how to adapt.
Written for developers already working with LLMs, Engineering in Flow goes beyond prompts and demos to cover what actually works in real systems:
- Designing single- and multi-agent architectures
- Retrieval-augmented generation (RAG) that survives production
- Human-in-the-loop patterns for oversight and control
- Guardrails, validation, and hallucination mitigation
- Observability, testing, and debugging for non-deterministic systems
- Scaling from fast prototypes to durable infrastructure
Rather than treating AI as a replacement for engineers, this book frames models as high-leverage collaborators-powerful, fallible, and requiring deliberate constraints. You’ll learn how to engineer flow as a repeatable development mode, not a lucky accident, while avoiding the chaos that unchecked automation creates.
Grounded in real deployments and written with technical precision, Engineering in Flow is for software engineers, AI practitioners, and technical leaders who want to ship faster without surrendering correctness, safety, or craft.
