Operating Model in the Age of AI: Five Shifts
B · Useful, limited A short consultancy perspective with useful structure and named case examples, but almost no original data beyond one Accenture survey figure and a client case with estimated numbers.
Summary based on the full text·Original: about 5 minutes to read
In short
A September 2026 Accenture perspective by Kent McMillan and Sanam Gill (Accenture) with Myles Kirby and Tom Oliver (Faculty, acquired by Accenture in 2026). It argues that once every company has the same models, the operating model becomes the limit on AI returns. It proposes five shifts: decisions as the unit of value, cross-functional value streams, people and agents managed as one workforce, cost per outcome instead of headcount budgets, and learning loops built from decision history.
Key numbers
- 58%of 3,000 C-suite leaders plan to change how work is split between people and AI in the next 12 months
- <10%of users and workflows drove the vast majority of AI spending in an Accenture deployment study
- 70% / 39%touchless cash processing and estimated capacity freed in one client's lead-to-cash redesign
As reported by the source.
English·Web page·opens accenture.com
What it reports
- Accenture's July 2026 Pulse of Change survey (n=3,000 C-suite leaders): 58% plan to change the human-AI split of work within 12 months, more than double the share at the start of 2026.
- An Accenture study of AI deployments found fewer than 10% of users and workflows drove the vast majority of AI spend.
- A global industrial solutions client redesigning lead-to-cash reached 70% touchless cash processing, with an estimated 39% of capacity freed.
- Amazon packages handled per employee rose from about 175 (2015) to 3,870 (2024); more than 1 million robots by 2025; its DeepFleet model improved robot travel efficiency by 10%.
- Mastercard's fraud system assesses about 159 billion transactions a year; generative AI doubled the speed of detecting compromised cards.
- Faculty's NHS England early warning system forecast hospital bed capacity about three weeks ahead during Covid-19.
- Moderna merged HR and IT under one Chief People and Digital Technology Officer in 2025.
- A 90-day test per shift, for example a live decision register, agents listed in the workforce view with named owners, and a cost-per-outcome baseline for one value stream.
How they got it
Consulting synthesis drawing on Accenture client work, one Accenture survey and public company material (Amazon, Mastercard, Moderna via press and company releases). No new dataset or method is described.
Limits
- Case numbers are client or company claims; the 39% capacity figure is an estimate.
- No evidence is offered that firms making these shifts outperform those that do not.
- The authors sell operating-model redesign.
Why it matters for companies
The practical point is about planning systems: payroll and contractor budgets have no line for agents, whose costs behave like cloud consumption and concentrate in a few users. That makes cost per outcome hard to see without finance, HR and technology sharing one view. The 90-day signals are a cheap way to check whether an AI program is changing how work runs or only adding tools. Giving every agent a named human owner is a low-cost governance step available now.
Summary based on the full text. Read the original for the full argument and data.Read the original on accenture.com