Deloitte: Only 15 Percent of Companies Can Already Coordinate Multiple Agents at Scale

On August 12, Deloitte published research based on a survey of 501 executives at U.S. organizations involved in autonomous-AI strategy or implementation, together with interviews with twenty managers and specialists. The results show a rapid shift from experimentation to operations, but also a large gap between scaling individual agents and creating a genuinely coordinated autonomous enterprise. Forty-two percent of respondents are still testing a small number of systems or have only a few production deployments. Another 43 percent are already expanding agents across multiple functions. Only 15 percent, however, have reached a stage where several autonomous systems operate in a coordinated way at scale. The practical difference is substantial: one application may, for example, draft a response to a customer, while a coordinated system must connect customer service, billing, inventory, shipping, and possibly refunds. Most companies therefore face a choice between simply adding new systems to an existing process and genuinely redesigning the process. Deloitte reports that 74 percent of leaders expect that within four years nearly half of business processes will be redesigned around agents. Sixty-one percent expect most processes to operate continuously with agent support and to enable real-time decision-making. The same 61 percent expect agents to be predominantly autonomous, while people move mainly into supervisory roles. Shorter-term expectations are much lower. Only 31 percent of organizations expect such extensive redesign within two years. In the same period, only a quarter of respondents expect coordination of agents across several functions, but within four years that figure rises to 58 percent. These numbers show that companies do not see agentic systems as a one-off software project, but as a multi-year change to the operating model. The shift will also affect work. Forty-three percent of leaders expect significant to extreme job disruption within 12 to 18 months, and 72 percent expect that level of impact within two to three years. Respondents expect routine and structured tasks to become increasingly autonomous, while creative, strategic, and highly judgment-intensive work remains under stronger human supervision. For leadership, it is important not to confuse the number of agents with maturity. Ten isolated automations may increase local productivity while also creating ten new handoff points and ten different sets of rules. Deloitte recommends starting with the desired business outcome and only then connecting use cases, data, governance, technical architecture, work design, and workforce changes. An autonomous enterprise will not emerge simply by placing another layer of software on top of an old process. Terms to explain Multi-agent system: An environment in which several autonomous programs cooperate and hand tasks to one another in pursuit of a common objective. Operating model: The way a company allocates processes, roles, decision rights, technologies, and responsibilities. Orchestration: Coordinated management of several systems so that their linked activities produce a single outcome.

Deloitte: Only 15 Percent of Companies Can Already Coordinate Multiple Agents at Scale

The full article is available to users registered at Hard Skills