Follow ZDNET: Add us as a preferred source on Google.
ZDNET’s key takeaways
- Scaling the AI agents in business is now a focus on accountability and governance.
- Half of working hours may be reshaped by the use of AI agents.
- Business accountability for AI agents will require humans “in the lead’ versus “in the loop.”
In 2025, agentic AI was still mostly a promise. Today, it’s a stress test for all companies across all industries.
The results of three separate research studies — from Deloitte, KMPG PwC, and Accenture — all point to the same uncomfortable truth: Companies are moving fast on adoption and much slower on the harder work of actually rebuilding how they operate with humans and agents as part of their labor force.
Deloitte’s Agentic Transformation survey found that while 43% of organizations are now expanding AI agent deployments across functions, only 15% have reached scaled, orchestrated multi-agent deployments. Workforce readiness sat at just 20%, and only 16% of businesses said their current processes were actually prepared for agentic adoption. That’s a big difference from where leaders think they’re headed. Deloitte’s report found that 74% of leaders expect half of business processes to be redesigned around AI agents by 2030.
Also: Businesses must reinvent their processes and workforce to scale agentic AI adoption
New research from two other major firms that are guiding large-scale implementations of AI agents fills the picture, and the consensus is remarkably consistent: adoption of AI agents is real, but production-grade, value-generating deployments are still rare.
That said, research from Salesforce shows that the number of active AI agents in organizations has tripled over the last year, and that AI agents have improved their capabilities by 350%, enabling them to handle complex tasks. More interestingly, employee use of AI agents has also increased threefold as trust deepens. The research found that the average number of agents per organization nearly tripled (from 5 to 13), while creation time dropped by 53%, to an average of 1.9 days per agent.
Shift from deployment to accountability
KPMG’s Global AI Pulse, based on a survey of 2,145 C-suite and business leaders across 20 countries, shows organizations moving from experimentation toward broader deployment of AI agents. But the center of gravity is shifting from deployment to accountability, AI economics, and value. The shift is due to the fact that the return on investment from AI agent adoption remains limited even as adoption climbs.
Also: Business adoption of AI agents tripled this year – as measurable ROI emerges
KPMG’s report found that 76% of businesses now see real business value from AI, a 12% increase in one quarter. In addition, 78% of business leaders are confident they can future-proof their AI strategy, up 8% since Q1 2026. Seventy-one percent of organizations say they are making good progress toward a fully integrated AI-human workforce, up 11% in one quarter. Organizations with full visibility into AI operating costs are five times more likely to report established ROI than those without such visibility.
Adoption of AI agents is accelerating, but the data shows the barriers are also accelerating, including difficulty scaling use cases and skill gaps, which have each roughly doubled quarter after quarter as the top obstacles to demonstrating ROI.
The report’s core findings are that the differentiator between companies that are pulling ahead with AI agent deployments and those that are stuck is not how many agents they’ve deployed, but whether they have clear accountability, stronger governance, and real visibility into what running AI at scale actually costs.
Also: 12 rules of agentic AI for successful enterprise transformation
Salesforce research reveals the importance of leadership as companies transition to becoming agentic businesses. More than two-thirds of middle managers are optimistic about AI’s role in the future of work, and they feel personally accountable for their team’s adoption of AI tools. Salesforce research also shows that most AI pilots focus on capability and speed — and skip the hard work of earning trust from the business. The 12 rules of agentic business transformation highlight what companies are doing to successfully scale their agentic AI production deployments.
The accountability challenge
A joint Accenture-Wharton study, building on Bureau of Labor Statistics task-level data across 18 industries, noted that “intelligence may be scalable, but accountability is not.” The research found that 50% of working hours across the US economy, including 120 million workers, are now being reshaped by roughly 60 digital and physical AI agents. In banking and capital markets specifically, digital AI agents alone touch more than 45% of hours worked.
Also: ‘Specialists aren’t required’ anymore: How to stay valuable in an AI agent workplace today
Modeling a hypothetical $60 billion company, the research forecasted roughly $6 billion in potential revenue growth and $1.7 billion in annual productivity gains from agentic AI at full maturity. The report found that across use cases, leading organizations are moving beyond isolated solutions and instead relying on a coordinated set of digital and physical AI agents that operate under human direction.
The Accenture report found that AI agents are spreading across enterprise systems faster than formal governance strategies can keep up. Accenture’s James Crowley, a co-author of the report, noted: “We like to say humans in the lead, not in the loop.” The distinction is deliberate because the human “in the loop” infers merely a human reviewing what the agent did, versus a human “in the lead” means that the accountability for the work to be done is with the human, not the agent. This shift raises a new leadership mandate: redeploy expanded capacity into measurable value and sustained growth.
The Accenture report also cautioned that productivity gains only become growth if leaders deliberately deploy freed-up capacity toward higher-value work; otherwise, productivity gains stall at efficiency and fail to translate into growth.
Also: Why replacing staff with AI backfires – and 5 ways smart leaders generate real value instead
Accenture’s proposals include a new business role (chief agentic resource officer), explicit P&L targets, human-led operating models, and clear decision rights defined before agents ever go live, not after.
Salesforce research showed that 70% of companies deploying customer service AI agents see ROI in 60 days. Agentic AI adoption for service organizations has grown from 39% to 66% in the past 12 months. The accountability challenge can be met with new outcome-based pricing models that focus on explicitly tying business outcomes to AI agent execution.
It’s more about relational transformation
The research from Deloitte, Accenture, and KPMG tells a single, coherent story that should reframe how leaders talk about agentic AI. AI adoption is not the hard part. Most companies have agents live; in fact, adoption has increased threefold in the past 12 months.
The hard part, as Deloitte’s data on workforce and process readiness first suggested, is everything adoption exposes: governance frameworks are needed for autonomous AI agents, the gap between deployment and demonstrable ROI, the gap between piloting and true broad usage, and the leadership discipline required to convert efficiency into growth without losing accountability.
Also: The 3 types of people who will excel in the AI agent era, according to tech leaders
None of these firms is arguing against agentic AI. They’re converging on a more precise argument: the technology has arrived faster than the operating model, the governance model, or the workforce readiness needed to run it responsibility at scale.
The next 12 to 24 months will separate the organizations that treat agentic AI adoption at scale as an integration challenge from those that correctly treat it as a leadership challenge and opportunity. Companies will need strong human relationships in order to weather the ambiguity and friction of change.
Business leaders will need strong human-AI relationships in order to ensure that digital labor becomes a source of leverage rather than confusion, passivity, or mistrust. They will also need to think far more seriously about how relationships between systems and agents are structured, governed, and monitored.
Becoming an agentic business is less about technology transformation and more about relational transformation.
