Insights
AI-Native Delivery: the 2026 field guide
AI-native delivery means running program and portfolio delivery with agentic AI doing part of the work, under human accountability. This hub pulls together what that looks like in practice: the operating model, the governance that survives an audit, and how to assure programs where agents contribute to the outcome.
Where the market actually is
Agents are deployed, and the return has not followed at the enterprise level. McKinsey's 2026 State of AI survey (1,719 respondents, May to June 2026) finds eight in ten people reporting that AI improved their own productivity, while the share of organisations reporting any impact on earnings is unchanged from a year earlier at 37%, and the share it calls high performers is about 6%. The strongest correlate with enterprise impact in that survey is fundamental workflow redesign, not the tools.
The forecasts point the same way. Gartner projected in October 2024 that 33% of enterprise software applications will include agentic AI by 2028, up from less than 1% in 2024; in June 2025 it forecast that more than 40% of agentic AI projects will be cancelled by the end of 2027, for escalating costs, unclear business value or inadequate risk controls. PMI's 2026 Pulse of the Profession puts AI and automation at the top of what is driving project complexity, named by 72% of respondents, ahead of market pressures at 48%, and finds that 31% of complex projects now miss the full scope of their intended benefits.
The constraint, in other words, is no longer capability. It is governance, verification and change: who owns the agent's output, how it is checked, and what happens when it is wrong. DORA's research on the return from AI-assisted development is organised around the initial "productivity dip" of a rollout, and finds that AI acts as an amplifier of whatever an organisation already is. That is the gap this guide addresses, from a delivery-assurance point of view rather than a tooling one; the evidence is set out in full in The Permanent Pilot, by the Numbers.
Sources as of September 2026: McKinsey, The State of AI: Global Survey 2026; Gartner press releases of 21 October 2024 and 25 June 2025; PMI, Pulse of the Profession 2026 (16th edition, 12 May 2026); DORA, ROI of AI-assisted Software Development and the 2025 State of AI-assisted Software Development. The McKinsey, Gartner, PMI and DORA wording was verified against the publishers' own pages on 11 September 2026. Figures are attributed to those sources and are not proprietary Zenous data.
Read the cluster
Eight pieces that together describe AI-native delivery end to end
McKinsey, PMI, Gartner and DORA, read together: why AI programmes stall, and the three things the assurance evidence says to fix. Read →
How leading portfolios assign agents to functions on the org chart, with owners, authority, and escalation. Read →
What the PMO function looks like when agent-assisted analysis and reporting become the default. Read →
A practical governance model for AI in delivery: what agents may touch, and how output is verified. Read →
The new audit checklist for programs where autonomous agents contribute to delivery. Read →
The warning signs that an AI transformation is a demo dressed up as delivery. Read →
Why the time from question to decision is the metric AI-native delivery should optimise. Read →
A grounded look at where AI tooling earns its place in project management, and where it does not. Read →
Delivery intelligence, occasionally
A low-volume note when we publish something worth your time: a new field guide, a sharp delivery idea, no filler. Unsubscribe anytime.
Put agentic AI to work inside delivery, safely
We help delivery organisations decide where agents fit, what they may touch, and how their output gets verified, with accountability for the outcome.