Rewiring Software Delivery - Second Edition (October 2026)
AI is changing software delivery faster than most enterprise operating models can adapt.
Coding agents can inspect repositories, generate implementations, run tests, invoke tools, investigate failures, and hand work to other agents. But faster execution creates a harder leadership problem: who authorised the work, what was the agent allowed to change, which context was trustworthy, what evidence proves the result, and when should autonomy stop?
Rewiring Software Delivery introduces the Agentic Software Delivery Operating Model, or ASDOM, a practitioner operating model for redesigning enterprise software delivery around human-agent execution rather than simply adding AI tools to the existing software development lifecycle.
Version 2 substantially rebuilds the book around ASDOM 2.0 as an explicit operating model. It separates organisational authority from technical capability, introduces distinct governance lanes for delivery agents and AI-enabled products, and follows one connected NovaBuild change from intent and architecture through execution, validation, security, release, operations, measurement, recovery, and institutionalisation.
The agentic-delivery platform is also expanded in Version 2. It now covers governed context, orchestration topology, tool and skill contracts, durable execution, evaluation, shared control planes, bounded operations, economics, readiness, recovery, and a three-phase transformation path from controlled proof to repeatable scale.
The book introduces several practical innovations for enterprise agentic delivery. These include bounded authority grants, governed context records, delegation controls, durable execution rules, independent validation, incident-authority ladders, evidence-driven release decisions, and explicit restoration of authority after failure.
A continuous NovaBuild evidence chain connects decisions across the delivery lifecycle instead of treating each chapter as a separate example. The Five-Minute Evidence Chain gives leaders a compact way to reconstruct the path from approved intent to architecture, authority, context, execution, validation, release, operational evidence, and restoration.
ASDOM also separates external law, standards, and established engineering practice from author-proposed operating-model constructs and illustrative case evidence. This makes it clearer what is externally established, what is an ASDOM design recommendation, and what is being used purely to demonstrate the method.
The result is not another book about using AI coding assistants. It is a playbook for technology leaders who need to redesign how software delivery works when more execution can be delegated to agents but authority, evidence, accountability, risk, and business outcomes still belong to people.
It is written for CTOs, CIOs, engineering leaders, enterprise architects, delivery leaders, platform leaders, transformation leaders, and senior practitioners responsible for bringing agentic AI into real enterprise software delivery.
