What agentic AI delivers with the right foundation
Designing Agentic AI for Results
15 Apr, 2026
Why agentic AI ROI requires a different kind of strategy
Most organizations have explored AI agents on a surface level. Pilots that showed promise. Demos that impressed the leadership team. Agents they’re considering.
What most organizations have not done is built a design strategy. That distinction costs more than they may realize.
Organizations are treating agentic AI like a technology decision when it is actually a strategic and design discipline.
Designing for returns
I have been working in data for over a decade, and I see the same pattern: organizations are treating agentic AI like a technology decision when it is actually a strategic and design discipline.
It requires a more considered approach: looking hard at your technology platforms, your data, your processes, and your organization, and building a real plan to make all of those components ready to support autonomous action. That work is what determines whether an agent delivers.
Forrester's Total Economic Impact study of Microsoft's agentic AI solutions found that organizations deploying agentic AI on a properly built foundation realized $44.5 million in benefits over three years against $20.2 million in costs, producing an ROI of approximately 120%.¹ This is what’s possible when leaders treat agentic AI as a design discipline from day one.
Most organizations are skipping that work and then wondering why the returns are not showing up. The research tells you exactly why.
Agentic AI, by Design
Discover the art of acceleration. 85% of organizations increased AI investment last year. Only 6% saw any return (Deloitte, 2025). The difference isn't the technology. It's the design. HSO turns agentic AI potential into measurable enterprise outcomes faster, with less risk, and with results that speak for themselves.
Start with the outcome
The most common mistake I see in agentic AI implementations is surprisingly simple.
Clients come to us and say: Here is how I do this specific process today, step by step. Please automate it exactly like that. That thinking produces agents that replicate existing workflows. It does not produce outcomes.
This is where design thinking comes into play. You move from initial idea to a visual representation of it, put it in front of real users immediately, gather feedback, refine it into a prototype, validate again, and arrive at a version one design. The whole cycle can run in days. What it produces is not a faster version of the old process. It is a process redesigned around the outcome you actually need.
Results Across Industries
15,000
hours saved every year
Retail Distribution
98%
reduction in manual processing
Hospitality
40,000
applications processed in week one
Finserv
8 weeks
kickoff to launch
Public Sector
Digging into data
The single most common reason AI initiatives stall is not the technology. It is the data.
Out-of-the-box large language models are trained on general data. The quality, availability, and structure of your data determines almost everything that comes after. Organizations that already have strong data management in place, typically those in heavily regulated industries that are accustomed to stricter guidelines, are moving faster and seeing better returns than everyone else.
If your data is not ready, your agents will not be either.
Three elements that determine whether your investment pays off
Design for the outcome, not the process. Most organizations come to us wanting to automate exactly what they do today. The ones that see real returns work with us to define what success looks like at the end, then design backward from there. That is design thinking in practice.
Get your data in order. The better your data is in terms of quality, availability, and structure, the better your agents will perform. This is the foundation everything else depends on. It is also the mistake most organizations discover too late.
Set the right expectations from day one. AI is non-deterministic, meaning it does not give the same output every time the way traditional software does. That is the point. But it means guardrails and governance have to be designed in, not added after. Do not over-promise. Build trust incrementally. Let the results speak.
The organizations that close the ROI gap in the next 12 to 18 months will have done it by committing to a real strategy, designing for outcomes from the start, getting their foundations right, and working with a partner who knows how to move from idea to operating reality.
When the design is right, the results will follow. That’s how you accelerate.
Sources:
Forrester Consulting. The Total Economic Impact of Microsoft's Agentic AI Solutions. Commissioned by Microsoft. Interviews with representatives across six organizations, survey of 420 respondents.
Alex Zweekhorst
Director Data & AI
Alex Zweekhorst is Director of Data and AI at HSO International and Global Data Practice Lead, helping enterprise organizations across manufacturing, retail, distribution, and professional services build the data foundations and AI strategies that deliver measurable outcomes.
Design Your Results
Ready to move from agentic AI potential to real payoff? Let's design your path forward.