Maria Mingallon

Maria Mingallon, Knowledge and Information Manager and AI Lead for Asia-Pacific at Mott MacDonald, brings a rare hybrid engineer-coder perspective to the AI adoption conversation. With a civil engineering background (she designed bridges), a Master's in emerging technology and design, and coding skills dating back to university (Java, Visual Basic), she has been automating engineering calculations since approximately 2008. The conversation digs into what "real" AI adoption actually means, AI Forum NZ reports adoption rising from 67% to 82% (Sept 2024 to March 2025), but just downloading ChatGPT does not count. Real adoption requires governance frameworks, use case identification, skills development beyond prompting, information security, and understanding AI architecture. She details Mott MacDonald's structured Copilot M365 rollout: targeted licenses (not everyone), employees must apply with a specific use case, mandatory knowledge shares every two weeks, and licenses removed if not used.

Key Topics Discussed

• AI adoption stats. AI Forum NZ reports 67% to 82% adoption (Sept 2024 to March 2025), but "what counts as adoption?" Just downloading ChatGPT is not real adoption
• Real AI adoption requires. governance/risk management, use case identification and prioritisation, skills development (beyond prompting), information security, understanding AI architecture
• Knowledge base quality. "garbage in, garbage out"; bias exists in every model (ChatGPT, DeepSeek, Claude Sonnet)
• Mott MacDonald's Copilot M365 rollout. targeted rollout (not everyone), employees must apply with specific use case, mandatory knowledge shares every 2 weeks, info security and governance training, regular surveys, licenses removed if not used
• Governance as key pillar. can't have staff putting confidential project info into free ChatGPT
• Tool overwhelm. people overwhelmed by number of AI tools; need to focus on few that align with organisational principles; 3-6 months to mass adoption but hundreds of new tools by then
• Small business vs enterprise. Andy says "if my team don't use enterprise ChatGPT, they're fired" (joke/serious) vs Mott MacDonald's structured framework
• Engineer-coder hybrid skillset. growing community of engineers who can code; Maria learned Java and Visual Basic at university, has been automating since ~2008
• Humanoid robots. Andy wants one but worried about the neighbours
• Andy and Python. Andy teaches himself Python, automates boring tasks with AI; Maria has been doing the same since 2008
• Career background. civil engineering (bridges), Master's in emerging technology and design at a prestigious European architectural school

Notable Quotes

"What counts as adoption?" Just downloading ChatGPT does not equal real adoption.

"Garbage in, garbage out." Knowledge base quality is critical for AI outputs.

"If my team don't use enterprise ChatGPT, they're fired.", Andy (joke/serious) on small business AI adoption.

Guest Background

Maria Mingallon is the Knowledge and Information Manager and AI Lead for the Asia-Pacific region at Mott MacDonald. She has a civil engineering background (designed bridges) and a Master's in emerging technology and design from a prestigious European architectural school. She learned coding (Java, Visual Basic) at university and has been automating engineering calculations since approximately 2008, a pioneer in the hybrid engineer-coder skillset.