The Emergence of Agentic AI
AI agents can increasingly execute workflows proactively and autonomously, creating potential to make organizations smaller and more efficient.
3.9 years
expected time to materialize
What does this trend encompass?
AI agents are becoming more proactive and autonomous, with their capabilities expanding to generate chain-of-thought reasoning and perform structured, goal-directed reasoning. Agents are becoming increasingly capable of planning, deploying, and executing complex business workflows across multiple functions without human intervention, conducting transactions and managing resources. If governance and accountability challenges can be solved, this could enable organizations to become more efficient and faster decision-making. As a subset of the 2025 trend on Deploying Autonomous Systems at Scale, this trend focuses specifically on systems cognitive autonomy.
Why is it important?
Since AI agents are trained on and embed powerful domain knowledge, they can transform businesses and create significant disruption, as well as prompting the need for adaptation. Unless access is widened, their transformative benefits may concentrate in large enterprises. Building trust in AI agents will demand robust oversight, accountability frameworks, and human-centered mechanisms to ensure they operate reliably and align with organizational and societal values.
Enabling conditions and countries’ readiness
As highlighted by DET survey respondents, preparing for the materialization of this trend at the country level depends on the following key drivers:
Digital innovation: while R&D and innovation drive the development of agentic AI, an important consideration in this process is the conversion of workflows into blueprints for agents that incorporate safety tests.
Digital infrastructure: scalable compute is needed in the form of servers, data centers, and high-speed communication networks so multiple agents can run reliably without overloading core systems.
Digital Capabilities: workers will need to be (re)trained to have the capability to collaborate with, provide clear workflow instructions to, and effectively supervise agents’ actions.
Industry Digital Transformation: industries will need to revise and adapt existing workflows to define permissible, industry-specific sets of actions for agents to take while creating intermittent checkpoints and final, key handoffs where human review will be needed.
Digital Policy and Governance: regulation must focus on setting guardrails around agent actions, provenance, incident reporting, and minimum safety tests to ensure that agent actions remain auditable and accountable across workflow processes.
Impacts on the horizon
Prospective turning points that could catalyze this trend into rapid, widespread materialization
Production guardrails for AI agents
become default as major providers include built-in policy checks, and human-review paths, standardizing and safe-guarding agent workflows
Cross-app execution normalizes,
allowing agents to carry out tasks across diverse functions such as resource planning, customer relationship management, and software delivery under one verified identity
Agent-run transactions become mainstream
as payment systems and machine identities mature into an integrated ecosystem, allowing agents to place (low-risk) orders and reconcile invoices under spend limits and audit trails.
Recommendations
Private sector
Pilot agentic AI in high-volume digital workflows
such as customer inquiries, invoice matching, and order tracking, establishing clear spending limits, approval authorities, and human review triggers for autonomous decision-making.
Co-develop an "AgentOps" governance layer with industry peers
that standardizes identity verification, policy adherence checks, budget caps, and audit trails to enable traceable, accountable autonomous operations.
Plan for interconnected agentic AI ecosystems
by developing interoperability norms and defining protocols that specify when a human review is required versus when AI agents can proceed independently within companies and entire value-chains
Retrain workers to supervise agent-led processes,
focusing on reviewing agent decisions, mapping workflows into agent blueprints with clear goals and safety tests, and establishing productive human-agent collaboration rhythms.
Public sector
Establish role-based liability frameworks that clearly assign responsibility:
data holders accountable for data quality and misuse, model providers for system behavior under stated uses, deployers for proper configuration and monitoring, operators for oversight and escalation.
Launch regulatory sandboxes where organizations can test agentic systems
conducting end-to-end digital transactions (processing claims, running procurement, placing orders) under supervised conditions with clear audit trails and spending limits.
Require minimum safety standards
including guardrails around agent actions, provenance tracking, incident reporting protocols, and human-in-the-loop review paths before agents can execute financial or legally binding transactions.
Co-invest in digital skills initiatives
to develop a workforce equipped to collaborate, supervise, and co-create with Agentic AI systems through, for example, prompt engineering, agent orchestration and AI ethics.
IGOs, IOs, and others
Convene a global forum for agentic interoperability
to develop common standards for agent identity, cross-application execution, and machine-to-machine transactions that enable agents to operate across different platforms and jurisdictions.
Create an independent certification system
evaluating agentic AI on decision transparency, data provenance, policy adherence, and audit trail completeness – providing clear trust signals for consumers and businesses navigating autonomous digital services.
Develop open-source agent blueprint libraries and workflow templates
that smaller organizations can adapt, lowering barriers to deploying cognitive autonomy and preventing concentration among tech giants with proprietary systems.
Fund demonstration projects
showing how agentic AI can automate bureaucratic processes, improve public service delivery, and reduce administrative burdens in emerging economies through transparent, accountable autonomous operations.
Read the Digital Economy Trends 2026 report
Explore the full insights and analysis of the 2026 research.