Capgemini Research: Trust and Human-AI Collaboration Key to $450B Agentic AI Potential

Agentic AI is forecast to generate up to $450 billion in economic value by 2028, yet only 2% of organizations have fully scaled deployments, and trust in autonomous AI agents is declining. According to Capgemini Research Institute’s latest report, “Rise of Agentic AI: How Trust Is the Key to Human-AI Collaboration”, organizations are finding that AI agents deliver the most value when humans remain actively involved.

The study reveals that confidence in fully autonomous agents has dropped from 43% to 27% over the past year amid growing concerns around data privacy, explainability, and ethics. Nevertheless, human-AI collaboration is expected to redefine work: within three years, AI agents are projected to be involved in most business tasks, with 65% more human engagement in high-value activities where collaboration is effective.“The economic potential of AI agents is significant, but realizing this value depends on more than just the technology,” said Franck Greverie, Chief Portfolio & Technology Officer, Capgemini. “It demands responsible development, strategic transformation, and the creation of human-AI chemistry to enhance judgment and outcomes.”

Transparency, Trust, and Teaming Will Define Agentic AI’s Success

Despite strong executive interest—93% say AI agents offer a competitive edge—only 14% of companies have started implementation, while nearly half lack a clear adoption strategy. Trust remains a major barrier:

• Only 40% of organizations trust AI agents to operate autonomously.

• Two in five leaders believe the risks outweigh the benefits.

• Trust is higher (47%) among those already in the implementation phase, versus 37% in early exploration.

As a result, organizations are prioritizing transparency, ethical safeguards, and decision-making explainability to bridge the trust gap and increase adoption.

Agentic AI is evolving from a tool to a team member. More than 60% of organizations expect to form human-agent teams in the next year, positioning AI agents as collaborators rather than just assistants. Furthermore, 70% believe AI agents will trigger organizational restructuring, requiring new workflows and redefined roles.

Effective collaboration is projected to lead to:

• 65% increase in human engagement in critical tasks

• 53% increase in creativity

• 49% improvement in employee satisfaction

Scaling AI Agents Could Unlock Massive Business Value

The Capgemini report identifies a $450 billion opportunity by 2028, split between revenue gains and cost savings from scaled deployment of semi- and fully-autonomous agents. Organizations with scaled implementations stand to earn $382 million on average over the next three years, compared to $76 million for less mature adopters.

Initial deployments are concentrated in customer service, IT, and sales, with future expansion into operations, R&D, and marketing. Yet, only 15% of business processes today operate with any meaningful level of AI autonomy—a number expected to grow to 25% by 2028."AI agents are still primarily copilots, not commanders," the report notes, "supporting routine tasks rather than independently running workflows."

Organizational Readiness Still Lagging Behind AI Ambition

Despite the hype, most organizations lack the infrastructure and preparedness to scale agentic AI:

• 80% lack mature AI systems

• Fewer than 1 in 5 have high data readiness

• 51% cite privacy as a top concern, yet only 34% have taken concrete steps to address it

Understanding is also limited: only half of business leaders say they truly understand AI agent capabilities or can identify areas where agents outperform traditional automation.

Capgemini’s recommendation? Move beyond experimentation. Organizations must:

• Redesign workflows around AI-first principles

• Strategically blend agent autonomy with human oversight

• Build transparency and ethics into every layer of AI development

As trust emerges as the defining lever, human-AI collaboration—not full automation—will drive the agentic AI era forward.