All insights
Fintech AI ConsultingSeptember 2, 20266 min read

The Rise of Agentic AI: Transforming Modern Business and Society

Agents now run service calls, back offices and, in the UAE, half of government by mandate. What that does to work, and who is accountable when software acts.

In this article
The Rise of Agentic AI: Transforming Modern Business and Society

The first wave of generative AI answered questions. The second wave does the work. An agent is a system that can take a goal, break it into steps, use tools and systems to carry them out, and keep going until the job is done or a person tells it to stop. In 2026 that shift moved from demos into operations, and it is changing how businesses are staffed, how governments deliver services, and what regulators expect of both.

This piece is not about agents paying for things, which we have covered elsewhere. It is about agents running the back office and the front line: the invoices, the service calls, the filings and the approvals that make up most of what organisations actually do.

Where adoption really stands

McKinsey’s 2026 State of AI survey is the cleanest measure. Nearly nine in ten organisations now use AI regularly in at least one function, and 44% say it is scaling across the enterprise, up from 38% a year earlier. Agents are where the gap opens. Among organisations with more than $1 billion in revenue, 40% report scaling AI agents in at least one function, up from 27% the year before. Among smaller organisations the figure stayed flat at 22%.

Two details in that survey say more than the headline. About one in five organisations is already scaling software-coding agents, and 32% have decided against buying at least one software product because an agent let them build it in-house. Agents are not only automating work; they are beginning to change what companies buy. The sobering counterweight is that only 37% of respondents can tie AI to any effect on earnings, which is the same proportion as last year.

What agents are doing in operations

The McKinsey Global Institute’s 2026 study of AI-embedded workflows describes a customer operations deployment that is becoming typical. Before, the company’s automated phone system resolved about 10% of enquiries and handed the rest to people. After deploying conversational agents across its customer base, agents handle roughly 40% of all calls and resolve more than 80% of those without human involvement. Cost per call fell by about half, and customer satisfaction rose six points because waiting times shrank and handling became consistent. Human representatives moved to the complex, sensitive and high-value cases.

The pattern repeats in sales, where the same study found time savings of 30–50% across roles, and in software, where agents take on most repetitive execution while people shift to planning, orchestration and testing. In financial operations the equivalent is reconciliation, document handling, first-line compliance checks and report drafting, where the work is structured, high-volume and already measured.

Agents do not remove the human from the process. They move the human to the exceptions, and make the exceptions the job.

What it does to work

The honest answer on jobs is “more reshaped than replaced, but not evenly.” BCG’s 2026 analysis of 1,500 US occupations finds that 57% of jobs depend too heavily on physical presence or sustained human interaction to be disrupted soon. Of the remaining 43%, most will stay but evolve, 14% will need significant upskilling, and about 12% face net losses where demand is capped and the agent substitutes directly, call-centre representatives and certain financial-analyst roles among them.

The pressure lands first at the bottom of the ladder. PwC’s 2026 AI Jobs Barometer finds early-career postings have flatlined in the most AI-exposed sectors, while junior roles in those sectors are seven times more likely to demand senior skills such as judgement and strategic thinking; “seniorised” entry-level roles have grown 35% since 2019. The World Economic Forum’s June 2026 report on entry-level work adds the worker’s view: 68% of entry-level staff say AI has raised their productivity, 45% say they now work longer hours, and 28% expect half or fewer of their current skills to still matter in three years.

The same research is clear about the upside. PwC finds productivity growth 40% higher at the companies most exposed to AI. The MGI study estimates that more than 70% of today’s skills apply equally to automatable and non-automatable work; people spend less time preparing documents and more time framing questions and interpreting results.

The Gulf is treating it as policy, not technology

Nowhere is the shift more deliberate than the UAE. On 23 April 2026 the UAE Cabinet announced a framework to move 50% of federal government sectors, services and operations onto agentic AI within two years, with ministers and directors-general to be assessed on how fast they adopt it. Eleven days later Dubai extended the programme to its private sector through the Dubai Chamber, with training tracks, incubators and investment vehicles. In June the Cabinet approved a single Federal Authority for Artificial Intelligence and Data to own policy, legislation and data governance.

Business is already ahead of the mandate. KPMG’s 2026 UAE Tech Report finds 97% of UAE organisations embedding AI agents into workflows, products or services, against 87% globally, and Abu Dhabi’s government has more than 100 AI use cases in production across more than 40 entities under an AED 13 billion digital strategy. Data residency is part of the design: the G42–Microsoft partnership announced in July 2026 exists to let governments and businesses deploy agents while keeping sensitive data inside the country.

The governance gap

Adoption is running ahead of accountability. A 2026 Gartner survey of 297 chief human resources officers found that only 24% of leadership teams agree on who is accountable for AI agents, and Deloitte reports that fewer than half of organisations piloting agents have defined a human–agent operating model. Regulation is arriving to force the question. The EU AI Act’s high-risk obligations took effect in August 2026, and read differently for agents than for models: human oversight has to sit in the action path as an approval gate on consequential steps, logging has to capture the full execution trace, and accuracy has to be measured continuously because an agent’s behaviour drifts. In the UAE, the Central Bank’s February 2026 guidance applies the same logic to financial institutions, with a consumer’s right to human review of automated decisions.

What leaders should do now

  1. Pick workflows with a number attached. Cost per call, days to close, exceptions per thousand. The deployments that pay are the ones that were measured before the agent arrived.
  2. Design the exception path before the automation. Who the agent escalates to, with what context, is the product. The MGI case worked because people took the hard 20%, not because the agent took the easy 80%.
  3. Name an owner for every agent. Three in four leadership teams have not. Accountability, permissions and a kill switch are governance, not paperwork.
  4. Redesign entry-level roles rather than cutting them. The WEF finds companies with the strongest AI returns are twice as likely to redesign workflows. Apprenticeship on exceptions is how the next generation of judgement gets built.
  5. Build for the regulator you will have in two years. Full execution logs, approval gates and continuous evaluation satisfy the EU AI Act, the CBUAE guidance and the UAE’s new federal authority at once.

The bottom line

Agentic AI is transforming business and society in a specific, measurable way: it is absorbing structured work, concentrating human effort on judgement and exceptions, and forcing organisations and governments to decide who is accountable when software acts. The Gulf has chosen to make that transition a mandate with a deadline. Everyone else will get there by competition instead. Either way, the organisations that will be comfortable in 2028 are the ones that built the exception paths, the governance and the training before the agents arrived.

How relevant and useful is this article for you?
From Ripae

Ripae Labs builds applied AI for credit, fraud and operations, with privacy and auditability designed in from the first line.