For years, employees were told that artificial intelligence would make work easier. It would automate repetitive admin, speed up workflows, improve productivity, and free people to focus on more meaningful work.
That first phase has already arrived.
Across industries, employees now use AI to draft emails, summarise meetings, generate reports, analyse spreadsheets, answer customer queries, and automate routine tasks. Companies invested heavily. Executives expected rapid productivity gains.
The gains were more uneven than expected.
According to a new report by the Top Employers Institute, fewer than 40% of organisations reported meaningful productivity benefits from early AI adoption.
The Institute, which certifies and researches workplace practices globally, drew its findings from data gathered from almost 2 500 certified Top Employers, alongside external research and organisational case studies.
Its latest report, Designing work for the age of agentic AI, examines how AI is moving beyond isolated tools and beginning to reshape workflows, organisational structures, accountability, performance measurement, and the role humans play inside companies.
The report argues that many businesses introduced AI tools without redesigning how work itself was organised. The result was fragmented efficiency gains, uneven adoption, confusion around accountability, and growing mistrust in some workplaces.
The next phase of AI, however, looks very different.
And for employees trying to understand what their future role inside a company may look like, the report offers some revealing clues.
AI is moving beyond tools – and into workflows
The report focuses on “agentic AI” – systems made up of autonomous AI agents capable of reasoning, making decisions, and acting with limited human supervision to achieve specific goals.
Earlier workplace AI largely functioned as an assistant. Employees still controlled the workflow, while AI supported isolated tasks.
Agentic AI begins changing the workflow itself.
In multi-agent systems, several AI agents co-ordinate together across processes that would previously have required multiple employees or departments.
The report describes systems capable of co-ordinating recruitment, onboarding, training, customer service interactions, supply chains, scheduling, operational support, and pricing decisions with minimal human intervention.
That shift, the Institute argues, changes “the structure of work itself”.
The significance of that line becomes clearer later in the report.
As intelligent systems become more embedded in workflows, employees increasingly move away from repetitive execution work and towards supervision, judgement, validation, exception handling, and co-ordination across systems.
In other words, AI is not simply changing tools employees use; it is beginning to change what companies expect employees to contribute.
The future employee may look very different
One of the report’s more revealing concepts is what it calls the “AI Orchestrator”. This refers to employees who use AI systems to operate with the output and reach of what previously may have required a small team.
The report does not predict mass job losses. In fact, human resources executives surveyed by Salesforce expect that 61% of workers will remain in their current roles while working alongside agentic AI systems.
But the report strongly suggests that the nature of those roles is beginning to shift.
Traditional assumptions around fixed responsibilities, specialised functions, clear departmental boundaries, and stable career structures may become harder to maintain as intelligent systems take over larger portions of workflow co-ordination.
The value employees bring may increasingly lie less in producing volume output themselves and more in:
- applying judgment
- interpreting context
- monitoring systems
- making decisions
- identifying risks
- and co-ordinating outcomes across functions
That appears to be where the report sees human contribution moving.
Vodacom’s case study offers a glimpse of where this is heading
The report includes a significant case study on Vodacom Group, offering a practical example of how a major local company is already embedding agentic AI across customer operations, HR systems, and internal workflows while trying to keep human oversight in place.
Its AI systems screen candidates, identify best-fit hires, answer employee policy questions, process requests, and flag engagement risks before they escalate into staff attrition.
According to the report, Vodacom’s AI-enabled recruitment systems have already reduced time-to-hire by 50%.
The company is also using agentic AI to optimise pricing, identify revenue opportunities, execute targeted campaigns, and process customer transaction reversals with limited human intervention.
What stands out most, however, is not only the technology itself; it is the company’s focus on training employees to work with it.
Vodacom has trained 531 employees through its internal “Citizen” programme to build automation and AI solutions inside the business. Those employees have already developed 658 automated and data-driven processes.
That may be one of the clearest indicators in the report about where organisations believe future value will come from.
The employees gaining importance inside these systems are not necessarily those resisting AI adoption – but those learning how to direct, manage, refine, and work alongside intelligent systems effectively.
Anxiety inside organisations is already growing
The report also makes clear that employee unease around AI is rising.
Although 96% of senior leaders remain confident that AI will improve performance, nearly half of employees say they do not understand how those gains are supposed to happen in practice.
The institute also cites growing concerns around:
- burnout linked to AI workloads,
- declining trust,
- fears around job security,
- surveillance discomfort,
- and the growing use of unauthorised AI tools inside organisations.
The report suggests that many organisations remain structurally unprepared for the scale of workplace redesign that may follow.
Only 61% of organisations surveyed have formal strategic workforce planning processes for future workforce and skills needs.
Only 41% use multiple future scenarios when planning workforce requirements.
That uncertainty may partly explain why the report repeatedly returns to themes such as trust, human oversight, ethical governance, and employee involvement in AI deployment.
The safest employees may become the most adaptable ones
One of the report’s clearest underlying messages is that organisations are starting to value a different mix of skills.
Technical AI capability matters. The report cites PwC research showing that workers with AI skills and experience are already commanding wage premiums of as much as 56%. But the report also suggests that human judgement is becoming more valuable, not less.
As AI systems become more autonomous, companies still need people capable of understanding nuance, applying oversight, making judgement calls, and maintaining accountability inside increasingly complex systems.
The report ultimately presents AI less as a story about human replacement and more as a story about organisational redesign.
Workflows are changing. Roles are stretching. Traditional boundaries inside organisations are weakening. Employees are not disappearing from that picture. But the definition of what makes an employee valuable appears to be changing very quickly.
