CEO Briefing: AI Agents Are Doing Actual Work.

AI agents have officially crossed the threshold from experimental tools to business-critical assets.

Recent breakthrough testing shows that AI agents can now complete complex expert-level tasks that previously took human professionals 4-7 hours, with some agents working autonomously for up to 30 hours on complex coding projects.

This represents a fundamental shift from AI as a support tool to AI as an autonomous workforce member.

OpenAI's latest testing reveals that advanced AI models are now performing at near-human expert levels across industries including finance, law, retail, and research.

Meanwhile, Anthropic's Claude Sonnet 4.5 has achieved 77.2% success rates on real-world software engineering benchmarks, and New Zealand businesses are already experiencing this transformation firsthand, 82% now use AI in some capacity, with 93% reporting improved worker productivity.

The key distinction is that these aren't just better chatbots, they're autonomous agents capable of multi-step reasoning, tool usage, and self-correction. They can replicate complex academic research, generate comprehensive business documents, and manage entire workflows without human intervention.

Why This Matters

New Zealand is experiencing an AI productivity boom that's outpacing global averages. 70% of Kiwi CEOs report AI has made their workforce more efficient, compared to just 42% in Australia. This isn't theoretical, it's happening now across sectors from manufacturing (Fisher & Paykel's Agentforce implementation) to finance (where AI agents monitor transactions 24/7).

The competitive implications are stark. Companies leveraging AI agents are becoming "Frontier Firms" that outperform competitors and expand workforce capacity without proportional hiring. With New Zealand's tight labour market, AI agents offer a strategic solution to talent shortages while maintaining our productivity edge.

Regulatory clarity also favours early movers. New Zealand's "light touch, proportionate and risk-based" approach creates space for innovation while existing privacy and governance frameworks provide adequate oversight. This regulatory environment gives Kiwi businesses first-mover advantages over competitors in more restrictive jurisdictions.

The Opportunity

The opportunity is transformational across three dimensions:

Operational Excellence: AI agents can automate entire business processes, not just individual tasks. They're handling customer service inquiries, processing complex financial analysis, conducting research, and managing supply chain logistics autonomously. Companies report 74% reduction in manual audit efforts and 63% cut in compliance-related downtime.

Strategic Advantage: Early adopters are creating competitive moats through AI-powered capabilities that competitors can't easily replicate. These aren't just efficiency gains, they're entirely new operating models that fundamentally change cost structures and service delivery.

Innovation Catalyst: AI agents excel at complex problem-solving and creative tasks. They can conduct comprehensive market research, generate strategic insights, and even accelerate product development cycles. The technology is enabling companies to tackle challenges that were previously resource-prohibitive.

The Impact on Risk and Governance

The autonomous nature of AI agents introduces unprecedented governance challenges that require board-level attention. When systems can take thousands of actions daily without human review, traditional compliance approaches become insufficient.

Operational Risks multiply exponentially as agents can initiate actions across multiple business functions simultaneously. Reputational risks escalate when AI interacts directly with customers and partners without human oversight. Financial risks compound when systems can commit organizational resources autonomously.

New Zealand faces specific cybersecurity concerns, with 28% of large businesses worried about AI-generated attacks, though only 6% of breaches have involved AI to date. Privacy risks are particularly acute given our robust Privacy Act 2020 requirements.

The interconnected nature of modern business amplifies these risks, an AI agent's decision in one area can cascade through supply chains, partner networks, and customer relationships in unpredictable ways. However, 69% of NZ executives see agentic AI as essential for coping with talent shortages and productivity pressures.

Questions To Ask

1. "What business processes could we completely reimagine if AI agents handled 60% of the workload?" Move beyond task automation to process transformation.

2. "How do we establish governance frameworks for systems that operate autonomously across multiple functions?" Traditional approval workflows are insufficient for real-time AI decisions.

3. "What new competitive advantages could we build through AI agent capabilities that competitors can't easily replicate?" Focus on strategic differentiation, not just efficiency.

4. "How do we balance the productivity gains from AI agents with the governance and risk management requirements?" Develop embedded compliance that builds regulatory requirements directly into AI operations.

5. "What workforce implications arise when AI agents become permanent team members rather than tools?" Plan for hybrid human-agent teams and new skill requirements.

6. "How do we prepare for the next wave of AI capabilities while maximising current opportunities?" Establish foundations for continuous AI evolution.

Next Steps

Immediate Actions (Next 30 Days):

• Conduct an AI audit of current usage across your organization, many employees are already using AI tools informally

• Establish an AI governance council with cross-functional leadership including legal, IT, HR, and key business units

• Assess your data infrastructure and security posture for AI agent integration

• Begin pilot programs with AI agents in low-risk, high-value processes

Medium-term Priorities (Next 6 Months):

• Develop comprehensive AI policies aligned with New Zealand's Public Service AI Framework principles

• Invest in workforce training for human-AI collaboration and establish new performance metrics for hybrid teams

• Partner with AI vendors to implement enterprise-grade agent solutions with proper security and governance controls

• Create measurement frameworks to track AI agent performance, risks, and business outcomes

The window for strategic AI agent adoption is now. Companies that move decisively will establish competitive advantages that become increasingly difficult for others to replicate.

The technology is ready, the regulatory environment is supportive, and New Zealand businesses are already demonstrating global leadership in AI productivity gains.

Research:

1. https://www.reuters.com/business/retail-consumer/anthropic-launches-claude-45-touts-better-abilities-targets-business-customers-2025-09-29/

2. http://journal.yiigle.com/LinkIn.do?linkin_type=DOI&DOI=10.3760/cma.j.cn112137-20240202-00252

3. https://arxiv.org/abs/2410.13639

4. https://www.anthropic.com/news/claude-sonnet-4-5

5. https://newzealand.ai/insights/ai-at-work-key-insights-from-new-zealands-latest-workforce-reports

6. https://www.linkedin.com/pulse/generative-agentic-ai-new-zealands-non-tech-sectors-jim-taylor-il2lc

7. https://kinetics.co.nz/ai-driven-productivity-gains-in-new-zealand-2025/

8. https://digital.nemko.com/regulations/new-zealand-ai-regulation

9. https://www.simpsongrierson.com/insights-news/legal-updates/ai-regulation-nz-walks-a-tightrope

10. https://www.dlapiper.com/en/insights/publications/2025/02/new-zealands-public-service-ai-framework-guiding-responsible-innovation

11. https://www.mckinsey.com/capabilities/mckinsey-digital/our-insights/superagency-in-the-workplace-empowering-people-to-unlock-ais-full-potential-at-work

12. https://newjaigs.com/index.php/JAIGS/article/view/366

13. https://www.mckinsey.com/capabilities/quantumblack/our-insights/seizing-the-agentic-ai-advantage

14. https://aws.amazon.com/blogs/aws/introducing-claude-sonnet-4-5-in-amazon-bedrock-anthropics-most-intelligent-model-best-for-coding-and-complex-agents/

15. https://www.nacdonline.org/all-governance/governance-resources/directorship-magazine/online-exclusives/2025/q3-2025/autonomous-artificial-intelligence-oversight/

16. https://www.insurancebusinessmag.com/nz/news/cyber/ai-sparks-new-fears-in-new-zealand-cybersecurity-sector-543750.aspx

17. https://www.privacy.org.nz/news/statements-media-releases/privacy-commissioner-speech-to-ema-about-ai/

18. https://www.ajg.co.nz/news/ai-cyber-attacks-key-concern-for-kiwi-businesses/

19. https://www.privacy.org.nz/assets/New-order/Resources-/Publications/Guidance-resources/AI-Guidance-Resources-/AI-and-the-Information-Privacy-Principles.pdf

20. https://chiefexecutive.net/the-ceos-guide-to-agentic-ai/

21. https://aiforum.org.nz/2025/02/25/ai-agents-mean-business-now/

22. https://www.mdpi.com/2073-431X/13/11/278

23. https://ieeexplore.ieee.org/document/10708696/

24. https://arxiv.org/abs/2410.23077

25. https://arxiv.org/abs/2409.18486

26. https://www.jmir.org/2024/1/e60807

27. https://arxiv.org/abs/2407.12126

28. https://ieeexplore.ieee.org/document/10801577/

29. http://biorxiv.org/lookup/doi/10.1101/2024.05.06.592800

30. http://arxiv.org/pdf/2410.05080.pdf

31. http://arxiv.org/pdf/2404.06411.pdf

32. https://arxiv.org/pdf/2502.00964.pdf

33. http://arxiv.org/pdf/2308.03688.pdf

34. https://arxiv.org/pdf/2409.15523.pdf

35. http://arxiv.org/pdf/2403.11905.pdf

36. http://arxiv.org/pdf/2411.04368.pdf

37. https://arxiv.org/pdf/2409.03215.pdf

38. https://arxiv.org/html/2411.01114

39. http://arxiv.org/pdf/2410.22553.pdf

40. https://insight7.io/writing-an-executive-summary-for-ceos-examples-and-tips/

41. https://www.xero.com/nz/guides/business-plan-exec-summary/

42. https://asana.com/resources/executive-summary-examples

43. https://visme.co/blog/how-to-write-an-executive-briefing/

44. https://www.indeed.com/career-advice/career-development/how-to-write-an-executive-summary

45. https://www.reddit.com/r/ArtificialInteligence/comments/1ff8mrk/open_ai_just_released_the_performance_of_their/

46. https://sdtimes.com/ai/anthropic-claims-its-newly-released-claude-sonnet-4-5-is-the-best-coding-model-in-the-world/

47. https://www.writingforcouncils.co.nz/how-to-write-an-executive-summary/

48. https://www.aisi.gov.uk/work/pre-deployment-evaluation-of-openais-o1-model

49. https://aws.amazon.com/blogs/aws-insights/the-rise-of-autonomous-agents-what-enterprise-leaders-need-to-know-about-the-next-wave-of-ai/

50. https://blog.hubspot.com/marketing/executive-summary-examples

51. https://time.com/7203729/ai-evaluations-safety/

52. https://simonwillison.net/2025/Sep/29/claude-sonnet-4-5/

53. https://www.domo.com/blog/autonomous-ai-agents-explained-what-they-are-and-why-they-matter

54. https://www.wgtn.ac.nz/__data/assets/pdf_file/0010/1779625/VBS-Report-Writing-Guide-2017.pdf

55. https://eu.36kr.com/en/p/3483351282539392

56. https://docs.claude.com/en/docs/about-claude/models/whats-new-sonnet-4-5

57. https://www.langchain.com/stateofaiagents

58. https://academic.oup.com/itnow/article/66/1/22/7609012

59. https://dl.acm.org/doi/10.1145/3643991.3644895

60. https://link.springer.com/10.1007/978-3-031-70074-3

61. https://prosiding.arimbi.or.id/index.php/PROSEMNASIMKB/article/view/20

62. https://revistas.unav.edu/index.php/communication-and-society/article/view/49904

63. https://www.mdpi.com/2413-4155/6/4/60

64. https://fcc08321-8158-469b-b54d-f591e0bd3df4.filesusr.com/ugd/185b0a_488292c2832b47baa9299ac1c3a82234.pdf

65. https://www.jier.org/index.php/journal/article/view/979

66. http://www.emerald.com/jsit/article/27/1/116-145/1239668

67. https://www.mdpi.com/2071-1050/16/3/1166

68. https://ijsra.net/sites/default/files/IJSRA-2024-0235.pdf

69. http://arxiv.org/pdf/2501.10388.pdf

70. https://linkinghub.elsevier.com/retrieve/pii/S0148296324000468

71. http://anapub.co.ke/journals/jebi/jebi_pdf/2023/jebi_volume_3-issue_1/JEBI202303005.pdf

72. https://www.mdpi.com/2673-2688/1/2/11/pdf

73. https://arxiv.org/pdf/2311.15473.pdf

74. https://www.mdpi.com/2071-1050/16/5/1790/pdf?version=1708581624

75. https://figshare.com/articles/journal_contribution/Advancing_innovation_in_financial_stability_A_comprehensive_review_of_AI_agent_frameworks_challenges_and_applications/28426736/1/files/52394846.pdf

76. https://arxiv.org/pdf/2402.17082.pdf

77. https://arxiv.org/pdf/2304.11771.pdf

78. https://kpmg.com/ch/en/insights/artificial-intelligence/governance-risk.html

79. https://fortune.com/2025/09/29/anthropic-releases-claude-sonnet-4-5-a-model-it-says-can-build-software-and-accomplish-business-tasks-autonomously/

80. https://www.spglobal.com/en/research-insights/special-reports/the-ai-governance-challenge

81. https://www.itnews.com.au/news/anthropic-launches-claude-45-touts-better-abilities-targets-business-customers-620661

82. https://gjeta.com/node/2028

83. https://www.scitepress.org/DigitalLibrary/Link.aspx?doi=10.5220/0012623000003645

84. https://www.emerald.com/jfra/article/23/4/1450/1253698/RegTech-advancements-a-comprehensive-review-of-its

85. https://www.tandfonline.com/doi/full/10.1080/13511610.2024.2349627

86. https://www.zf.uni-lj.si/images/zalozba/Sokratska_11/13_Ljube_David.pdf

87. https://ajee-journal.com/legal-analysis-of-eu-artificial-intelligence-act-2024-insights-from-personal-data-governance-and-health-policy

88. https://arxiv.org/abs/2409.14603

89. https://www.mao.kiev.ua/biblio/jscans/knit/2024-30/knit-2024-30-5-10-malysheva.pdf

90. https://hstalks.com/doi/10.69554/AYHZ2687/

91. https://policyreview.info/pdf/policyreview-2024-1-1746.pdf

92. https://arxiv.org/pdf/2207.01482.pdf

93. https://policyreview.info/pdf/policyreview-2024-3-1790.pdf

94. https://arxiv.org/html/2409.00015v1

95. http://arxiv.org/pdf/2306.12609.pdf

96. https://www.frontiersin.org/articles/10.3389/fhumd.2024.1421273/full

97. http://arxiv.org/pdf/2407.10305.pdf

98. https://arxiv.org/pdf/2503.05937.pdf

99. https://academic.oup.com/policyandsociety/advance-article-pdf/doi/10.1093/polsoc/puae020/57986099/puae020.pdf

100. http://arxiv.org/pdf/2407.16888.pdf

101. https://www.puttingdatatowork.com/post/agentic-ai-strategy-five-questions-every-ceo-should-be-asking-about-ai-agents

102. https://www.mbie.govt.nz/assets/responsible-ai-guidance-for-businesses.pdf

103. https://xmpro.com/the-ten-questions-engineering-executives-need-to-ask-when-considering-agentic-ai-for-operations/