Angus J. McLean is an AI director at OLIVER who applies artificial intelligence to advertising strategy, audience research, and creative production. His defining principle, bounded autonomy, holds that useful agents need carefully limited information, permissions, and responsibilities.
McLean studied cultural theory and worked as an events journalist before the pandemic prompted a career change. After retraining through online courses, he joined Ogilvy in December 2020 and became a senior social intelligence analyst, studying online conversations, audience behavior, and emerging trends. His client work included Dove, Nokia, IBM, Cargill, and the Gates Foundation; an IBM watsonx project examined how nonspecialists understood artificial intelligence. In 2024, he received Social Intelligence Insider 50 Rising Star recognition.
At OLIVER, he brought that cultural-analysis background to agent-assisted audience research, competitor analysis, localization, copywriting, and creative development. His approach to advertising agents centers on several practical convictions:
- Context engineering as behavioral control: Curated documentation and credible consumer material can guide an agent more effectively than unrestricted internet access, which often introduces promotional claims and search-optimized noise. Larger context windows make editorial selection more important, not less.
- Semantic clustering for cultural insight: McLean uses clustering, labeling, and structured analysis to turn sprawling conversations into material that creative teams can act on. One example organized approximately 50,000 social posts into themes, representative posts, and potential advertising strategies.
- Simple systems and short feedback loops: A résumé-building experiment convinced him that straightforward HTML could outperform an unnecessarily elaborate application. He favors smaller models, custom harnesses, preprocessing, and representations such as graphs, folders, and timelines when they solve a specific problem.
- Human judgment before automation: Agents inherit the weaknesses of poorly understood workflows. McLean insists that practitioners understand a task themselves before automating it, particularly when audience sentiment, irony, and brand risk resist literal interpretation.
His writing on ecology, intelligence, and biomimetic design extends these concerns beyond advertising, examining how cultural assumptions shape definitions of intelligence and what designers can learn from biological systems.
