Your organisation can buy better tools and fund more training and still hand people a working day with no room to use either. How the time saved by AI is spent and who pays for learning are decisions made at the top, and they decide whether today's delivery builds the capability you will need tomorrow.
Andy Demir | Qalbur Perspective | Future of Work | 13 minute read
Executive Summary
Most organisations run employee experience, learning, and productivity as three programmes with three owners, yet employees experience all three on the same working day. AI makes the connection hard to ignore: faster production can release capacity, shift effort into review or raise expectations, depending on choices few boards ever see being made. Drawing on earlier work on employee experience and learning organisations, and on current research, this essay argues that senior leaders should deliberately allocate productivity gains, fund the time people need to practise, and keep performance standards clear. The task at the top is to improve delivery without quietly dismantling the conditions that make people more capable.
In This Article
- The working day is where the promise is tested
- What the earlier books still help us see
- AI changes the composition of effort
- Learning access is an operating decision
- Account for the time that improvement releases
- Performance standards belong in the design
- What senior leaders should consider
- Questions for the Boardroom
- The organisation you are teaching people to become
- References
The working day is where the promise is tested
An employee can come back from an AI training session your company paid for, keen to try what they learnt, and walk into a day with no room for it. The deadlines, the approval chain and this month's number are all still there, so the employee now has to negotiate the cost of using what they learnt, usually in unfinished work or their own evenings. From the boardroom, meanwhile, the training budget looks spent and the programme delivered.
That gap is what this essay is about. Employee experience is shaped by the conditions under which people produce results, and those same conditions determine whether they can produce better results next year. My argument is that work design must connect current delivery with future competence, and that making this connection is a senior leadership responsibility. Without it, a company can keep improving its training offer while making learning steadily harder to practise.
Gallup's State of the Global Workplace 2026 report, based on data collected in 2025, puts global employee engagement at 20%, its lowest level since 2020. More telling for anyone running an organisation, manager engagement fell from 27% to 22% in a single year, and Gallup attributes most of the recent decline to managers. These figures rest on Gallup's own definition of engagement, so they do not mean that four in five employees are unwilling to work. They do mean you cannot read commitment from the fact that people are still on the payroll, and that the layer you rely on to turn strategy into a working day is the one losing heart fastest. [1]
So before you commission another survey, ask what actually happens when someone in your organisation tries to improve the way work is done. Who gives permission, which existing task disappears, and which performance measure changes? The answers show whether development is part of your operating model or an extra obligation imposed on the people who do the work.
What the earlier books still help us see
Jacob Morgan's The Employee Experience Advantage (Wiley, 2017) describes employee experience through three environments: physical, technological and cultural. Its lasting contribution is to make the work environment a management responsibility, and Morgan places the initiative with the chief executive and the executive team rather than leaving it to HR. A capable person still needs usable tools, a suitable place to work and relationships that make contribution worthwhile. [2]
Some of its enthusiasm needs care. A layout that costs less per desk can cost more in distractions, and coaching and acting on feedback consume management time even when no software is needed. Morgan's comparisons between companies also cannot show that spending on experience caused better results, because successful companies have more to spend in the first place.
Holly Burkett, a learning and performance specialist, supplies the complementary argument in Learning for the Long Run: 7 Practices for Sustaining a Resilient Learning Organisation, published by ATD Press in November 2016. She connects learning with strategy execution, measurement, change capability, and knowledge sharing, and she asks the question that matters most to a board: does learning survive when budgets tighten or the C-suite changes? [3]
A third book, The Future Workplace Experience by Jeanne Meister and Kevin J. Mulcahy, published by McGraw-Hill in October 2016 with a 2017 copyright, links workplace experience with technology, on-demand learning and a changing workforce. It is a useful bridge, though its forecasts were written a decade ago and need to be tested against 2026 before anyone treats them as a description of it. [4]
Read together, these books broaden what counts as a good working environment: conditions that help people deliver and opportunities to become more capable through delivery. A pleasant workplace with little development can become a comfortable dead end, and an ambitious learning programme within an overloaded one becomes just another source of pressure.
AI changes the composition of effort
The strongest case for AI at work deserves a fair hearing. In Microsoft's 2026 Work Trend Index, 58% of surveyed AI users said they were producing work they could not have produced a year earlier. The survey covered 20,000 knowledge workers who already use AI, across ten markets including France, Germany, Italy, the Netherlands and the UK, between February and April 2026. It is a vendor-sponsored survey of users, so it measures neither all workers nor productivity directly, but a self-reported expansion of capability on that scale still matters. [5]
The same research shows why that potential is hard to capture, and the cause sits at the top. Among the AI users surveyed, 45% said it felt safer to concentrate on current goals than to redesign their work with AI, and only 13% said they were rewarded for reinventing work even when results fell short. Your people are receiving two instructions at once: improve the system, and protect the output of the existing one. Microsoft's own analysis gives organisational factors such as culture, manager support, and talent practices more than twice the weight of individual mindset and behaviour, and no learning platform resolves the conflict created by the incentive system. [5]
The report also hints at where the effort goes. Asked which human skills matter most as AI takes on more work, half of the users named quality control of AI output, the top answer. A calculation that counts minutes saved on drafting can overlook the review, coordination, and exception handling that take their place, and you can test that with evidence from your own workflows. [5]
The broader European evidence provides a reason to consider control as well as convenience. A European Commission Joint Research Centre working paper published in July 2026, drawing on the EU-wide AIM-WORK survey, associates algorithmic management with lower autonomy, reduced ability to take breaks, and greater work-related stress. The associations vary by practice and by country, and they are strongest where algorithms direct how tasks are done and how fast. The paper is about algorithmic management, a narrower thing than every use of generative AI, and it reports associations without proving cause. [6]
For anyone signing off a technology investment, the difference is practical. A tool that helps someone prepare a customer response and a system that sets the pace of every response produce very different working days, which is why review capacity, recovery time and clear responsibility belong in the design alongside the automation.
Learning access is an operating decision
Burkett's concern with sustainable capability becomes more demanding when the working day changes this quickly. People need to practise on real tasks, know what good output looks like and get feedback from someone qualified to judge it. A completion rate tells you an activity took place, and nothing about whether the person can now do the work, yet it is often the figure a learning dashboard leads with.
The OECD's evidence points back at the employer. Its 2025 report Trends in Adult Learning, based on the 2023 Survey of Adult Skills, finds that most adult learning takes place at the workplace during working hours, that lack of time and cost are the barriers adults cite most often, and that employer support is one of the strongest predictors of whether people take part at all. Whether your people learn, in other words, depends heavily on conditions you control. [7]
The same survey shows how differently people learn across Europe. In Portugal, 41% of employed adults and in Spain, 35% said they learn new things at work every day, against 4% in Poland, and across the countries surveyed overall participation in adult learning ranges from 58% in Norway to 13% in Korea. These self-reported figures describe the kind of work people do and the opportunities around them as much as individual behaviour, so they are no ranking of national ambition. For a group operating across several countries, this is a warning that one course catalogue will not erase differences in education, work organisation, and available time. [8]
The design question is sharpest on the frontline. An office employee can move a meeting to practise with a new tool, while a technician or service worker usually needs cover, access to equipment or a scheduled handover. That difference is operational and has nothing to do with commitment. A training budget without money for that cover produces unequal access, even though it looks equal in the board pack.
Personalisation needs the same realism. An AI tutor can explain a procedure at a person's own pace, but its answers still need reliable source material and verification, and passing a quiz generated by the system that provided the explanation is a weak substitute for demonstrating the skill in real work.
Account for the time that improvement releases
My suggestion is to treat saved time as a resource that needs an explicit allocation decision. This is a proposal for Qalbur readers, untested as a formal model, and it starts from an ordinary observation: an efficiency gain has several possible destinations. It can go into more output, better quality, capability development, recovery or a lower cost base, and each choice has different consequences for your people and for the business. In most organisations, nobody makes that choice, and the gain is absorbed into higher targets by default.
Start with one workflow where the consequences are visible. In a commercial proposal, faster drafting only becomes a real saving after someone has checked the technical commitments, pricing assumptions, and the customer's requirements. Measure the whole process, including correction and the reviewer's time. If the gain disappears downstream, raising proposal volume only moves the bottleneck onto someone else's desk.
When capacity is released, decide how to use it before anyone raises a target. Part may go into more customer work and part into supervised practice or better response quality, and the split will look different in a cash-constrained business and a growing one. What matters is that the choice is made deliberately, written into the business case and revisited, so that the people who own the P&L can see it.
The same applies to learning time. A manager cannot protect practice if every existing commitment keeps equal priority, and that order of priorities is usually set above the manager. Say what will be delayed or covered, and accept a visible short-term cost where the expected gain in capability justifies it.
Measurement then becomes more useful. Track task quality, correction effort, time to competent independent performance and access across employee groups. That is a better basis for investment than logins or certificates, though a financial return on learning is still hard to isolate, because staffing, demand and process changes move the same numbers.
Performance standards belong in the design
There is a serious objection to all of this. A business cannot protect every preference, preserve every task or promise permanent employment. Customers expect reliable service, investors expect discipline, and some organisations have little room to carry spare capacity. An employee experience programme that avoids hard performance decisions shifts work onto the strongest colleagues and wears down their trust.
The objection sharpens the argument. Clear standards are part of a good experience, because people need to know what contribution is expected and how it will be judged, and investment in learning carries a reciprocal expectation that the skills will be used. Managers still have to deal with persistent underperformance, while trying to tell it apart from poor tools, inadequate instruction, or conflicting priorities.
Flexibility shows why evidence should discipline both sides of the debate. A randomised trial with 1,612 graduate employees at Trip.com, the Chinese travel technology company, found that working from home two days a week cut quit rates by a third with no detectable effect on performance reviews or promotions over the following two years. The 395 managers in the trial started out expecting a loss of productivity and ended up expecting a small gain. Published in Nature in 2024, the trial ran in 2021 and 2022 and excluded interns and staff on probation, so it supports that arrangement for that population and says little about frontline work, new starters or fully remote organisations. [9]
Location policy should therefore follow the task: mentoring and hands-on practice may require people to be together, concentrated work may benefit from flexibility, and people who cannot work remotely can still be offered predictability, a say in their schedules, and access to development.
The harder trade-off is which tasks automation removes. Routine work can be a burden, but it also gives beginners practice and gives experienced people stretches of lower mental demand. Taking it away creates a duty to redesign apprenticeship and workload. On balance, the case favours automating routine work and deliberately rebuilding the practice it used to provide, through supervised cases and progressively greater responsibility, so that you keep the efficiency gains without losing the route by which people learn judgement.
What senior leaders should consider
Start by making productivity gains a decision with an owner. When the executive committee approves an AI or automation investment, the business case should say where the released capacity will go, in what proportions, and who is accountable for checking a year later that it went there. Without that line, the default allocation is higher targets, and nobody at the top will have chosen it.
Then look at one material workflow end-to-end, with the people who do it, their manager, operations, and the owners of learning and technology, and ask where people spend effort compensating for the system. A pilot built on what you find needs explicit quality thresholds, named review ownership, protected time for practice and managers with authority to adjust priorities within agreed limits.
Put your managers' workload on the agenda. Gallup's figures make them the most exposed layer: their engagement is falling fastest, and a Gallup study of US managers finds that their engagement declines as spans of control widen. In its US data, Gallup also finds that employees whose managers actively support the team's use of AI are far more likely to say that AI has changed how work gets done. If you widen spans or add coaching duties without removing administration, the design is incomplete, and the cost will appear in the very layer your strategy depends on. [1]
Fund the cover that makes learning possible on the frontline, and put time to competent performance on the dashboard you actually review, next to output and cost. Finally, give someone the job of turning what people learn into better procedures and standards of judgement, so that the next person starts with a better system.
Questions for the Boardroom
- Where have we counted the time AI saves without counting the review and correction it creates?
- When we approve an automation business case, who decides where the released capacity goes, and who checks a year later?
- Which commitments are actually removed or covered when we expect people to learn, and who absorbs the cost?
- Which employee groups have access to training on paper but no realistic chance to practise in paid time?
- As routine work is automated, where will our next generation of experienced reviewers learn their judgement?
The organisation you are teaching people to become
Every working day teaches your people something about the company. It teaches them whether improvement is worth the effort, whether admitting uncertainty is safe and whether learning leads anywhere. Whatever leadership says is weighed against those accumulated lessons, including what happened the last time someone asked for time to practise.
Designing work around employee experience means taking responsibility for that education, and much of it is set by decisions that never feel like decisions about people: a business case, a target, a span of control. You still need results today, and some constraints will stay uncomfortable. But the way effort is allocated also decides who will be able to deliver tomorrow. The next time a productivity proposal reaches your table, look past the hours it promises to save and ask what capability your organisation will still have once it is implemented.
References
- Gallup (2026). State of the Global Workplace: 2026 Report. Gallup, Inc. Global engagement data collected January to December 2025; AI and manager-support findings based on US surveys.
- Morgan, J. (2017). The Employee Experience Advantage: How to Win the War for Talent by Giving Employees the Workspaces They Want, the Tools They Need, and a Culture They Can Celebrate. Wiley.
- Burkett, H. (2016). Learning for the Long Run: 7 Practices for Sustaining a Resilient Learning Organization. ATD Press. Published November 2016.
- Meister, J. and Mulcahy, K. J. (2016). The Future Workplace Experience: 10 Rules for Mastering Disruption in Recruiting and Engaging Employees. McGraw-Hill, first edition. Published October 2016; copyright 2017.
- Microsoft (2026). How Frontier Firms Are Rebuilding the Operating Model for the Age of AI. Work Trend Index findings, with survey methodology. Survey of 20,000 AI-using knowledge workers, February to April 2026.
- Mariscal de Gante Martín, A. and Villani, D. (2026). Algorithmic Management and Working Conditions in Europe: Evidence from the AIM-WORK Survey. European Commission Joint Research Centre, Working Papers Series on Labour, Education and Technology 2026/05, published 6 July 2026.
- OECD (2025). Trends in Adult Learning: Why Is Participation Not More Common? Based on the 2023 Survey of Adult Skills.
- OECD (2025). Trends in Adult Learning: Where, What and How Are Adults Learning? Based on the 2023 Survey of Adult Skills.
- Bloom, N., Han, R. and Liang, J. (2024). Hybrid Working from Home Improves Retention without Damaging Performance. Nature, 630, 920–925. Randomised trial conducted in 2021–2022.
About the Author
Andy Demir is an international commercial and business leader with more than two decades of experience across complex B2B and industrial markets. Andy writes about executive leadership, strategy, organisational transformation, commercial performance, international business and the changing world of work, combining research with lessons drawn from operating across industries, markets and cultures.
