James Barton argued in a recent article that many organisations say they want personalised, on-demand learning, but still keep humans involved at every stage of the process. That got us thinking about a related issue: at what point does human oversight stop improving quality and start simply duplicating work?
Sales enablement leaders are broadly aligned on where they want to go. They want learning that is personalised, available on demand, and relevant in the moment of need. They want content that adapts to the individual rather than forcing everyone through the same fixed experience.
The problem is that many organisations say yes to that future while still operating as if nothing has changed.
That is where adoption starts to stall.
The hesitation is often described as a quality concern. Teams want to keep humans involved. They want review, approval, and control. On the surface, that sounds sensible. In many cases, it is sensible.
But there is an important question underneath it: are we protecting quality, or are we protecting ourselves from change?
When organisations bring AI into enablement, the real challenge is rarely whether the technology can produce something useful. More often, the challenge is that old ways of working are being copied into a new environment. Instead of redesigning the workflow, teams simply add AI into the middle of an existing process and then keep every human step around it.
That does not create quality. It creates duplication.
In traditional content workflows, heavy human involvement made sense. Content took time to create. Editing cycles were slower. Personalisation at scale was expensive. Review processes were designed for a world where production capacity was limited and each piece of content carried a higher cost of change.
AI changes that equation.
It can generate first drafts faster. It can restructure information instantly. It can adapt content for different roles, regions, or scenarios without requiring the same manual effort every time.
Yet many teams still treat this new capability as though nothing fundamental has shifted. AI creates the first draft, but then the business routes it through the same layers of checking, rewriting, and approval that existed before. The result is a workflow that is technically modern but operationally old.
That is why so many AI programmes feel slower than they should. The technology moves quickly. The operating model does not.
This is where organisations need to be more precise.
Quality matters. In enablement, poor content can create confusion, reduce trust, and weaken seller performance. Human judgement still has an important role to play, especially where context, nuance, accountability, or brand risk are involved.
But not every human touchpoint improves quality.
Some steps genuinely strengthen the output. Others simply make the organisation feel safer because they preserve familiar roles, familiar controls, and familiar routines. That distinction matters.
If every stage requires human intervention by default, then the organisation has not really decided where human value sits. It has simply chosen to keep the old model intact.
That is not a quality strategy. It is a change avoidance strategy dressed up as governance.
A better question is not, “How do we keep humans involved everywhere?”
It is, “Where does human effort add the most value?”
That shifts the conversation from blanket control to deliberate design.
In practice, that means separating the parts of the workflow where AI can do the heavy lifting from the parts where people should apply judgement, experience, and accountability. It means designing for contribution, not comfort.
Once leaders start looking at the workflow in those terms, the trade-offs become clearer.
If teams routinely discard AI-generated drafts and recreate the work manually, the issue is not only output quality. It may also be a lack of trust, unclear standards, or poor workflow design.
If AI has accelerated creation but the approval process is unchanged, then speed gains are being cancelled out elsewhere. The workflow has expanded, not improved.
When organisations cannot clearly explain which decisions need human input and why, they tend to default to full human involvement at every stage.
Quality should be measurable and specific. If it is being used as a catch-all argument against change, it may be masking discomfort with the new way of working.
Organisations making real progress with AI are not the ones waiting for perfect outputs. They are the ones redesigning work around a more sensible division of labour.
They let AI handle more of the initial creation, structuring, and adaptation. They ask people to focus on the areas where judgement materially changes the result. They make oversight intentional rather than universal.
This leads to a different type of workflow.
The first draft, summary, adaptation, or content variant is generated quickly so teams are not spending time producing basic material from a blank page.
People step in where context, judgement, risk, or strategic relevance are critical. Their time is spent sharpening the output, not recreating it.
Instead of reviewing everything in the same way, organisations define what requires close scrutiny and what can move faster with lighter-touch governance.
Success is not judged by how many people touched the content. It is judged by relevance, speed, usability, adoption, and business impact.
This matters because enablement is under pressure from both sides.
On one side, the business expects faster, more personalised support for sellers. On the other, teams are being asked to do more without endlessly increasing headcount or complexity.
That tension cannot be solved by adding AI to an unchanged system.
If the goal is to deliver more relevant learning, faster content adaptation, and support at the moment of need, then the work itself has to be redesigned. Otherwise, AI becomes another layer in the process rather than a way of improving the process.
And when that happens, organisations often conclude that AI is overhyped, when the real issue is that they never changed the operating model around it.
The hardest part of AI adoption is not usually technical. It is organisational.
It requires leaders to decide where control is truly necessary, where trust can be built, and where familiar processes may no longer be serving the outcome they were designed to protect.
That is an uncomfortable decision because it affects more than systems. It affects habits, responsibilities, and identity. It asks people to stop equating involvement with value.
But that is also where progress starts.
The organisations that move fastest will not be those that remove people from the process entirely. They will be the ones that use people more deliberately.
The future of enablement is not human or AI.
It is knowing the difference between oversight and duplication, and designing work accordingly.