Building AI Products from Scratch— Volume 2: Prompt Engineering, RAG, Agents, and the Product Platform - Brossura

Devlin, Mira S.

Brossura

ISBN 13: 9798175356084

Casa editrice: Independently published, 2026

Riassunto

Building AI Products from Scratch—Volume 2
Prompt Engineering, RAG, Agents, and the Product Platform

A model provides intelligence. Engineering turns that intelligence into product capability.

Building an AI product requires far more than connecting an application to a model API. Production AI systems need structured context, reliable retrieval, controlled tools, durable workflows, backend services, user interfaces, and data systems working together as one platform.

Building AI Products from Scratch—Volume 2 moves from the foundations established in Volume 1 into the engineering of the intelligence and application platform behind modern AI products.

What You Will Learn

• Engineer production-grade prompts and structured context
• Design Retrieval-Augmented Generation (RAG) pipelines
• Build ingestion, chunking, embedding, indexing, retrieval, and reranking workflows
• Improve retrieval quality, relevance, freshness, and provenance
• Combine semantic, keyword, and hybrid search
• Connect AI systems to APIs, databases, applications, and external tools
• Design controlled and secure tool-calling architectures
• Understand agents, planning, execution, memory, and multi-step workflows
• Establish boundaries around agent authority and autonomous actions
• Design human-in-the-loop workflows for sensitive operations
• Build durable workflows capable of handling interruptions and failures
• Engineer backend services for AI workloads
• Design frontend experiences for streaming, feedback, and uncertainty
• Build data pipelines that support continuously changing AI applications

From Prompts to Production Capability

A prompt alone is not an AI architecture.

Production systems require deliberate control over the information provided to models and the actions that can follow their outputs.

This volume treats prompt engineering as software engineering. Prompts should be structured, versioned, evaluated, tested, and maintained. Context should be deliberately assembled rather than passed to a model without control.

RAG is treated as an information architecture, not a guarantee of factual accuracy. Retrieval quality depends on ingestion, chunking, indexing, search, ranking, metadata, access control, freshness, and provenance.

Agents introduce another critical engineering challenge. An agent may determine which actions are needed, but reasoning does not automatically create permission to execute them. Production agent systems require explicit tools, authentication, authorization, validation, policies, limits, monitoring, recovery, and human oversight where appropriate.

The central principle is:

Do not build an AI model into a product. Build a product system around AI intelligence.

Designed for AI Product Builders

This book is written for AI/ML engineers, software engineers, full-stack developers, system architects, platform engineers, technical product managers, engineering leaders, founders, and developers transitioning into AI engineering.

It is particularly useful for teams building AI assistants, enterprise copilots, RAG applications, intelligent automation, agentic systems, AI-powered SaaS products, knowledge systems, and AI-enabled business workflows.

The emphasis is on durable engineering principles and architectural reasoning rather than dependence on a particular model provider or framework.

Part of the Building AI Products from Scratch Series

If you want to move beyond basic prompts and AI demos and learn how to engineer retrieval, tools, agents, applications, and data into a cohesive AI product platform, Volume 2 takes you to the next stage.

Engineer the context. Build the retrieval system. Control the tools. Design the agents. Build the platform. Turn intelligence into capability.

Dettagli dell'edizione

• Editore: Independently published
• Data di pubblicazione: 2026
• Lingua: Inglese
• ISBN 13: 9798175356084
• Rilegatura: Copertina flessibile
• Numero di pagine: 350
• Contatto del produttore: Manufactured by Amazon on behalf of the author; https://www.amazon.it/hz/contact-us; ; c/o Amazon Media EU S.à.r.l., 38 Avenue John F. Kennedy; Luxembourg; L-1855; Lussemburgo