The future of work: How AI is reshaping skills, talent economics and organisational design

Artificial intelligence is changing not just how work gets done, but how organisations structure teams, develop talent and measure performance. The shift is already visible in India: Microsoft's 2026 Work Trend Index found that 32% of Indian AI users surveyed qualified as “Frontier Professionals”, employees actively redesigning work around AI agents, twice the global average of 16%.

The skills challenge is equally significant. PwC's 2026 Global AI Jobs Barometer found that skills required in the most AI-exposed jobs are changing more than twice as fast as those in the least exposed jobs. Meanwhile, McKinsey's The State of AI in 2026 found that 40% of respondents from large organisations reported scaling AI agents, up from 27% the previous year.

Together, these findings point to a fundamental shift: the next phase of AI adoption will depend not simply on what technology can automate, but on how effectively organisations redesign workflows, develop talent and measure business outcomes.

For Barry Fiske, Senior Vice-President and Chief Experience Officer of Merkle Americas, a dentsu company, the answer begins with rethinking how people learn and work. Navin Dhananjaya, Global CXM lead at Dentsu Global Services (DGS), emphasises the need to connect AI adoption with measurable business value.

In a conversation with Mint, the two executives discussed how organisations can prepare their people and processes for an increasingly AI-driven workplace.

From task automation to AI-led workflows

The next phase of enterprise AI adoption is about more than using tools to accelerate individual tasks. Agentic AI is enabling organisations to connect multiple steps of a workflow, allowing systems to execute tasks and coordinate activities with varying degrees of human supervision.

This is changing the boundaries between design, engineering, data and strategy. Dentsu refers to this convergence as “design engineering”, encouraging professionals to understand disciplines beyond their immediate specialisation.

“Our engineers have to understand aspects of design because agentic tools are allowing them to actually create things and design things,” Fiske said. “And our designers have to actually understand the language of engineering.”

However, broader capabilities should not come at the expense of specialist expertise.

“The core skills remain very, very, very important. But you have to have a broader spectrum,” he said.

The emerging requirement is for T-shaped professionals: people with deep expertise in one discipline and sufficient understanding of adjacent fields to collaborate effectively and use AI strategically. For employers, this means designing roles around outcomes and complementary skills rather than rigid task boundaries.

Rethinking the economics of talent

As AI takes on more execution, organisations will need to reconsider how they allocate talent, assess productivity and structure teams.

AI can enable individuals to accomplish work that previously required several specialists. Yet the economics of this shift are not determined by headcount or hours saved alone. Businesses must also account for inference and token costs, integration, oversight, and the rework required when outputs fall short.

The relevant question is whether AI improves the overall economics of delivering a product or service, not simply whether it makes one task faster.

Fiske cautions against assuming that AI-enabled individual contributors will necessarily outperform collaborative teams. Dentsu continues to value cross-functional collaboration, including work across geographies, where different perspectives can improve ideas.

The challenge is to determine which activities benefit from individual ownership and which require the combined judgement of specialists. AI can expand individual capacity, but it does not automatically replicate the creative value of diverse teams.

From human-in-the-loop to human-on-the-loop

As AI becomes embedded in workflows, organisations must decide how much human involvement each process requires.

A human-in-the-loop model places people directly within the decision or execution process. A human-on-the-loop model allows AI to perform tasks while people supervise, monitor performance, and intervene when necessary. More autonomous processes may be appropriate for certain bounded, low-risk activities, provided suitable safeguards are in place.

This is not an inevitable progression towards removing people. The appropriate model depends on the consequences of errors, the reliability of the system and the judgement required.

The longer-term opportunity lies in self-healing systems that can detect failures, diagnose issues and recover through predefined mechanisms. These capabilities could reduce manual intervention in routine technical workflows, but they require testing, monitoring and clear accountability.

As execution becomes increasingly automated, human expertise will remain essential to defining objectives, evaluating outcomes and determining when systems should be challenged or stopped.

From AI productivity to measurable business value

For Dhananjaya, the business case for AI must ultimately be reflected in financial performance.

“The benefit that we want to kind of track is if it lands up in a finance report. So, that's at least our goal,” he said.

The distinction matters as organisations move from experimentation to wider deployment. Faster coding, research or content generation may demonstrate operational improvement, but these gains need to translate into lower delivery costs, greater output, improved customer experiences or additional revenue.

Organisations must therefore measure AI performance across the entire workflow. Token budgets and computational costs need to be considered alongside employee time saved, output quality, error rates and human supervision.

Dhananjaya's emphasis on financial outcomes highlights the need to move beyond isolated pilots towards AI-native operating models, where technology is integrated into how work is designed, delivered and evaluated.

The challenge is ongoing. As AI capabilities evolve, businesses must continually reassess their workflows and operating assumptions rather than treating adoption as a one-time exercise.

Building organisations that learn continuously

Technology alone cannot deliver this transformation. Organisations must create environments in which employees can experiment, acquire new skills and apply them to real-world challenges.

Fiske describes this approach through the analogy of a “teaching hospital”.

“We like to talk about ourselves as a teaching hospital where we're actually producing things, but people are learning as they do it,” he said.

The idea is to combine learning with delivery, allowing employees to develop capabilities while working on live assignments rather than relying exclusively on formal training.

Dhananjaya cited Dentsu's Hot Chocolate initiative, through which new recruits are encouraged to bring fresh approaches to problem-solving. In one example, a market research team developed a gamified front-end and adapted descriptions of hot chocolate to appeal to Gen Z consumers.

Such experiments illustrate why AI-era talent development must extend beyond tool proficiency. Curiosity, problem-solving and the ability to connect insights across disciplines will become increasingly important.

Fiske also compares career development to writing a novel, with each chapter representing a new experience, challenge or opportunity to learn. Professionals who deliberately explore adjacent disciplines can build combinations of skills that help them see possibilities others may miss.

The next competitive advantage

The future of work will be shaped by the interaction between AI capability, organisational design and human expertise. Businesses that focus exclusively on automation risk overlooking talent development, collaboration, governance and the economics of implementation.

For professionals, the imperative is to preserve deep expertise while developing the broader capabilities needed to work effectively with AI and across disciplines.

For organisations, success will depend on building adaptive workflows, giving teams opportunities to learn and ensuring that productivity gains translate into measurable value.

The question is no longer simply how much work AI can perform. It is how organisations can redesign work so that technology and people together deliver better outcomes.

Note to the Reader: This article is part of Mint's promotional consumer connect initiative and is independently created by the brand. Mint assumes no editorial responsibility for the content.