When AI Expands the Role, Who Owns the Decision?
By Ann Harten and Craig Pattison FCIPD
For family-owned and privately held businesses, AI is changing what a single capable person can carry. That shift is already underway.
The early use cases may look modest: research, summaries, trend analysis, drafting, reporting and everyday productivity. Some businesses are experimenting. Others are still watching, unsure where the real value lies or how quickly to move.
But the deeper issue is not whether a business has started using AI tools. It is whether owners and boards have considered what AI may do to the role itself.
The first visible gains may come from faster tasks. The more significant gains will come when roles are redesigned around what AI now allows capable people to test, decide, escalate and own.
That is a different question from automation. It goes to the structure of work.
In the first article in this series, we argued that AI is putting the traditional organizational pyramid under pressure. In the second, we explored the emergence of a more diamond-shaped organization: a narrower base, a constrained top and a more consequential middle.
This third article looks inside that structure, at the roles that now have to carry more. If AI broadens what people and AI-enabled systems can do, what happens to handoffs, accountability and management judgment?
For family and private businesses, that is not just a role-design question. It is a stewardship question.
Why this matters more in family and private businesses
The logic applies to many organizations. But it has particular force in family-owned and privately held businesses.
In these businesses, judgment is often concentrated in fewer people. Decision rights may be shaped as much by trust, history and proximity as by formal governance. Institutional memory often sits with owners, founders, long-serving managers and trusted operators who understand the business in ways that are not always documented.
They know which customer relationship needs care, which supplier risk is real, which margin movement matters, which workaround is hiding a deeper problem, and which number does not feel right.
AI can support analysis. It does not preserve context or decision quality on its own.
That risk will not appear neatly in an AI business case. It appears later, when the business needs judgment it has not developed and finds the bench thinner than expected. That is why role redesign is a stewardship question, not only an efficiency one.
When one person carries the loop
For decades, work has often moved like a relay race.
Finance carried the numbers and passed them to legal. Legal reviewed risk and handed it to operations. Operations assessed delivery, then passed the work to the commercial team. Each function held its own piece of the truth, and consequential decisions often changed hands several times before anyone acted.
That model had value. Specialists improved quality. Boundaries created accountability. Reviews caught expensive mistakes before they happened.
But every handoff also carried a cost: translation, delay, dependency and loss of context.
Businesses accepted that cost because the alternative looked risky. No single person could know enough across finance, legal, operations, technology, customers and people to move too far on their own without creating exposure.
AI is challenging that assumption. It will not turn people into experts in everything. What it can do is give a capable person enough working command of the next domain in the flow to continue further than they could before.
As AI becomes more agentic, this becomes more significant. The tools are no longer only summarizing information or drafting outputs. AI-enabled systems can increasingly route tasks, trigger workflows, compare options, surface exceptions and prepare recommendations. That does not remove the need for human judgment. But it changes how much work can move before another person or function formally takes over.
A commercial leader may model margin exposure before finance formally reviews a deal. An operations manager may test resource scenarios before escalating a capacity issue. A product lead may explore technical constraints before involving engineering. A finance manager may interrogate operational data before asking for further explanation.
The specialist does not disappear. Their role becomes more targeted. They are no longer the gate every early idea must pass through. They become the judgment applied where the stakes, complexity or risk require it.
That can be a better use of expertise, but only if the organization is designed for it.
The accountability problem
If one person can carry a wider loop, authority and accountability must be redrawn.
Without that, family and private businesses create the appearance of speed while increasing unmanaged risk. People move faster, but no one is quite sure what they are allowed to decide, what must be escalated, or who owns the consequences when AI-supported work crosses functional boundaries.
A commercial leader may use AI-supported analysis to model a deal and move quickly toward approval. But if the margin assumptions have not been stress-tested against finance, cost volatility, rebate structures or delivery constraints, the risk may only become visible after signature.
This is where a form of shadow decisioning can emerge. The person appears to own the decision, but the options, assumptions or recommendation may already have been shaped by AI before proper human challenge has taken place.
That is the accountability problem. AI can help work move faster across functional boundaries, but unless decision rights move with it, the business may not know who owned the judgment at the point the decision was made.
This is especially important where formal governance and informal trust overlap. A long-serving commercial director may know when to call the owner. A finance lead may understand which margin movement matters. An operations manager may know which supplier issue can be managed locally and which one needs escalation.
Those systems are valuable. They are also vulnerable if AI changes the flow of work faster than the business updates its decision rights.
Speed without authority frustrates capable people. Speed without governance creates risk.
Broader roles require stronger judgment
AI extends reach. It does not supply wisdom.
A person using AI may be able to generate scenarios, compare contracts, summarize data, test assumptions or identify anomalies. But they still need judgment to know whether the output is credible, whether the data is incomplete, whether the recommendation is commercially sensible, and whether the issue carries reputational, legal, ethical or operational risk.
In a family or private business, those factors often include customer history, supplier trust, family intent, legacy commitments, community reputation, margin instinct and long-term stewardship.
The people best placed to benefit from AI are not the most enthusiastic users, but the ones who know when to trust it, when to challenge it, and when to bring in deeper experience.
This changes the role of managers. They become less like supervisors of activity and more like governors of decision quality. Their task is to test how decisions are being framed, what assumptions have been checked, where the limits of AI-supported analysis sit, and when specialist expertise is required.
That is a higher management standard than many businesses have yet designed for. Managers will need clearer decision rights, stronger commercial judgment, better AI literacy and the confidence to challenge work that looks complete but has not been properly tested.
Getting that right is what the operating model must now be designed around. Owners and boards should focus on three decisions in particular: which handoffs can safely go and which still protect the business; which roles should broaden and what authority must follow; and where AI is already shaping options, assumptions or recommendations before human review takes place.
The choice
The most dangerous response to AI is not caution. It is letting the operating model change by accident.
That happens when tools broaden roles, compress handoffs and shift accountability before owners and boards have decided what should change, what should be protected, and where human judgment needs to sit.
The opportunity is significant. A capable person, equipped with AI and working inside a system designed for it, can make better decisions sooner, with more options on the table and less time lost to unnecessary handoffs.
The businesses that get this right will not be those that moved fastest with tools. They will be those that were most deliberate about where human judgment still needs to sit — and built their operating model around that answer.
About the Authors
Ann Harten is the founder of C-Suite Advisor, an AI strategy advisory practice working with corporate enterprises on governance, operating model design, and leadership capability in the age of AI. She brings 25 years of executive experience spanning CIO, CHRO, and VP of Global AI Transformation roles, and completed MIT’s Executive Education program in Deploying AI for Strategic Impact. Ann is an NACD certified board director and serves on corporate boards.
Craig Pattison FCIPD Craig Pattison FCIPD is an executive coach and strategic advisor working with senior leaders, founders and leadership teams when progress has stalled and the usual conversations are no longer shifting the issue. A former Chief People Officer and Chief Technology Officer, he brings lived executive experience across listed, private equity-backed and regulated environments. He is the creator of Root-to-Result®, a proprietary leadership methodology for working back from visible outcomes to the deeper drivers underneath.
