Skip to main content
ADVERTISEMENT

Part 3 — Connecting Agents to Digital Ecosystems

Overview

An agent that cannot connect to the systems where work actually happens is a demo, not a deployment.

Part 3 addresses the gap between an agent that works in isolation and one that operates effectively inside a real organisational environment — integrating with legacy infrastructure, interacting with customers, navigating authentication and data pipelines, and surviving contact with conditions that no amount of testing fully anticipates.

This is the part of agentic deployment that is most frequently underestimated. The model capability question is largely solved. What remains hard is the connective tissue: trust design, integration strategy, and technical plumbing. Getting these right is less about AI than it is about engineering discipline, organisational context, and an honest understanding of the infrastructure most enterprises have actually built over the last three decades.

Chapter 9 starts with the human side — trust and customer-facing design, where the stakes for getting it wrong are highest and most visible. Chapter 10 examines the dominant integration challenge facing most enterprises: grafting intelligence onto systems that were never designed to accommodate it. Chapter 11 goes one level deeper into the technical realities of system integration — the problems that don't appear in architecture diagrams but dominate engineering calendars.


Chapters in This Part

ChapterTitleTheme
9Designing for Human Connection: Trust and Customer-Facing AITrust and UX design
10Grafting Intelligence onto Legacy: Integration StrategiesLegacy integration
11The Plumbing Problem: Technical Realities of System IntegrationTechnical architecture

Chapter 11 is the most technically detailed chapter in this part. Readers primarily interested in strategy and design may read Chapter 9 and Chapter 10 first.

Building agentic AI and wondering why alignment is harder than the technology? Get in touch

ADVERTISEMENT