From human in the loop to human above the loop
But once workflows truly change, the harder questions are about roles, incentives and decision rights.
When workflow change, structure have to follow
Think of AI not as a pilot, but as a fundamental redesign of how work gets done. Then a lot shifts:
- Systems gather information and make routine decisions within defined boundaries
- Reports, research and handovers largely run in the background
- Customer problems can be spotted before they escalate
- Many approval steps become exceptions instead of the rule
If the work changes this much, job descriptions, spans of control, KPIs and incentive systems no longer fit.
The gap: people learn, organizations stay the same
Organizations don’t work like technology stacks. They coordinate human effort. Yet in practice, people are trained to work differently while the organization around them stays largely unchanged.
The Microsoft 2026 Work Trend Index, based on a survey of 20,000 AI users in ten countries, shows how big the gap is: organizational factors such as culture, leadership support and talent practices have twice the impact on AI value as individual effort.
Incentives are a key reason. 45 % of AI users say it is safer to focus on current goals than to redesign their work with AI. Only 13 % say they are rewarded for reinventing work, even when the outcome is uncertain. Companies ask for new ways of working but reward the old ones. Training alone won’t fix that.
What changes for roles
Take management. Much of it today is coordination: collecting status reports, consolidating slides, scheduling follow-ups. When AI takes over this coordination, managers gain time for what only humans can do: setting goals, developing people, making hard trade-offs and building trust.
The same holds for most knowledge work. Jobs are bundles of activities such as research, synthesis, planning, analysis and stakeholder management. AI automates some, augments others and creates new ones. Over time, the bundle changes so much that the old job description no longer fits.
The World Economic Forum’s Future of Jobs Report 2025 shows the scale: employers expect core skills to change for 39 % of the workforce by 2030, and 63 % see skill gaps as the main barrier to transformation. It’s not only about technical skills: creative thinking, resilience, leadership and collaboration gain importance too.
A new division of labor: human above the loop
For years, “human in the loop” meant a person approves each step. In many cases that remains right. A second model is gaining ground: human above the loop. AI runs a well-defined process within agreed limits. People define goals, boundaries, quality standards and escalation rules, step in when something unusual happens, and stay accountable for the outcome.
That is different work, and it raises questions that are not IT questions:
- Measurement: How do we assess performance in a human-agent team?
- Accountability: Who is responsible for an AI agent, and who may change its decision rules?
- Capacity: What happens to the time that is freed up, and where are the gains reinvested?
- Development: How do juniors build expertise when learning tasks are automated?
- Governance: Where is human review valuable, and where is it just an expensive bottleneck?
- Incentives: Which new behaviors do we reward, and which metrics do we stop measuring?
Three takeaways
- Treat AI as an organization topic, not only a technology topic. Workflows, roles, incentives and decision rights need to be redesigned together.
- Align incentives with the ask. If you want people to rethink their work, make it safe and worthwhile to do so.
- Make the human question explicit. The question is no longer only “What can AI do?” but “What should humans do now?”
AI cannot answer these questions for a company. Leadership has to. It is less about becoming an AI engineer and more about making the organizational decisions that technology cannot make.
Author’s note: The observations in this article draw on recurring patterns from our work with organizations navigating AI transformation, combined with publicly available research. Client situations have been generalized and anonymized; no individual organization or engagement is being described.
Sources: Microsoft 2026 Work Trend Index; World Economic Forum Future of Jobs Report 2025; case studies on workflow and operating model redesign from McKinsey and BCG
