2025_AI_脉搏调研报告（第三卷）_从自动化迈向自主化_19页_1mb

AI Agents: From Automation to Autonomy Summary

Core Content

The document outlines the current and future state of AI agent adoption across various industries and organisational maturity levels. It highlights the rapid integration of agentic AI and multi-agent AI systems into core business operations and the strategic importance of balancing autonomy with human oversight.

Key Findings

• Current Adoption:

  • 23% of organisations have already integrated agentic AI into core operations.
  • 27% plan to do so within six months.
  • Semi-autonomous agents are the most preferred (38%), followed by context-aware assistants (25%), basic task automation (12%), and fully autonomous agents (19%).

• Maturity and Readiness:

  • Organisations in stages 4 and 5 (Optimisation and Transformation) are most prepared for agentic AI.
  • Stage 2 (Experimentation) and Stage 3 (Defined) are in the process of implementation, with slower execution.
  • Only 10% have achieved full transformation, and 7% are in the initial stage.

• Industry Trends:

  • Technology, healthcare, and financial services lead in agentic AI adoption, particularly in IT, fraud detection, and risk management.
  • Retail and manufacturing are catching up, focusing on operations and supply chain.
  • Government and education sectors are more cautious, with longer timelines and ongoing evaluations.

• Leadership Perspectives:

  • C-suite and board members are more open to full autonomy (37% for semi-autonomous, 31% for fully autonomous).
  • Operational teams are more cautious, with only 16% expecting full autonomy.

• Decision-Making Impact:

  • Agentic AI is expected to automate routine decisions and enhance real-time insights.
  • Stage 5 organisations are most optimistic about automation, with 77% expecting AI to take over repetitive tasks.
  • Healthcare and technology sectors show the strongest consensus on real-time decision-making improvements.

Integration Timelines

• Stage 5: 83% have already implemented or plan to do so within six months.
• Stage 4: 70% already integrated or planning within six months; 17% within 12 months.
• Stage 3: 29% plan integration within 12 months; 19% within one to two years.
• Stage 2: 36% within 12 months; 28% within one to two years.
• Stage 1: Over half expect integration within one to two years or three to five years.

Strategic Integration Approaches

Recommendations for Closing the Readiness Gap

1. Align autonomy goals with business strategy
   Ensure AI ambitions are tied to strategic objectives to avoid misallocation of resources.

2. Invest in scalable infrastructure and technologies
   Build resilient data pipelines, cloud platforms, and integration tools to support efficient AI operations.

3. Upskill and empower talent
   Focus on training, hiring, and enabling teams to work effectively with AI agents, including hybrid roles.

4. Determine readiness based on maturity level
   Use maturity models to evaluate governance, infrastructure, and ethical oversight, and set realistic timelines.

5. Develop ethical and responsible AI protocols
   Implement transparent and accountable frameworks to ensure AI operates securely and ethically.

6. Prioritise resilience over capability
   Emphasise building a robust foundation for AI systems, even if they are semi-autonomous, to ensure long-term success.

Survey Demographics and Methodology

• Survey Period: July 2025
• Participants: 899 respondents (n=899), including 171 board members and C-suite executives.
• Industries Represented:
  • Technology (14%)
  • Retail (11%)
  • Government and manufacturing (8% each)
• Geographic Distribution:
  • United States (39%)
  • India and United Kingdom (12% each)
  • Canada (9%)
  • Australia (7%)
• Organisational Functions:
  • IT (30%)
  • Operations (18%)
  • Finance and accounting (10%)
  • Human resources (8%)

Conclusion

The document underscores the growing adoption of agentic AI across industries, with a clear emphasis on balancing autonomy with oversight. It highlights the need for organisations to align AI strategies with business goals, invest in infrastructure and talent, and develop ethical frameworks. As AI maturity increases, so does the ambition for autonomy, but many organisations still face challenges in governance, data security, and ethical implementation. A balanced, strategic, and resilient approach is essential to successfully transition to autonomous AI systems.