Reframing agentic AI deployment as a strategic redesign rather than a software purchase

Agentic AI may be generating excitement in boardrooms, but the evidence increasingly suggests that enthusiasm alone does not produce returns. HSO’s Alex Zweekhorst argues that organisations are still treating agentic systems as a software buy, when they need to be approached as a design and operating-model challenge. That view is reinforced by Deloitte, which found that 85% of organisations increased AI investment in the past year, yet only a small minority are seeing rapid payback....

The case for a more deliberate approach is becoming stronger. A Forrester Total Economic Impact study commissioned by Microsoft found that, in retail and consumer goods, agentic AI deployments could deliver returns of 124% to 282% over three years, while another Forrester analysis cited by HSO put benefits at $44.5 million against costs of $20.2 million, or roughly 120% ROI, when organisations build on the right foundations. Microsoft says the bigger shift is not merely adopting agents, but rebuilding workflows, incentives and decision rights around them. (microsoft.com)

That is where many projects stumble. IT Pro reported that Forrester sees a familiar pattern: companies remain stuck in pilots because of orchestration problems, control issues and lack of trust, while fragmented data, registries and routing make it difficult to scale isolated agents. Deloitte similarly identified siloed platforms and weak data quality as major barriers, warning that proof-of-concepts often collapse once real operational data is introduced. (itpro.com)

HSO’s argument is that organisations should begin with the outcome they want, then work backwards to redesign the process, rather than automating today’s workflow step by step. The company also stresses that AI’s non-deterministic nature means governance, guardrails and expectations must be built in from the start. Its campaign materials point to reported results such as 15,000 hours saved annually in retail distribution, 98% less manual processing in hospitality, 40,000 applications handled in a first week in financial services, and an eight-week launch in the public sector. (crmsoftwareblog.com)

The broader strategic question is also changing. An arXiv paper published in June 2025 argues that agentic AI can move firms towards autonomous business models, in which AI begins to execute core value-creation and value-capture mechanisms rather than simply support them. Deloitte’s research points in a similar direction, saying the most successful organisations are treating AI as enterprise transformation, using different ROI frameworks for generative and agentic systems, and investing in operating-model change rather than one-off automation. (arxiv.org)

Source: Noah Wire Services