The AI Corner

Mike Bayly

Spotlighting Kiwis moving the local and global AI ecosystem forward.

  1. 1d ago

    Ipsos AI Monitor 2026: why 65% of Kiwis are nervous, only 21% trust AI outputs, and sentiment has not moved in three years | The AI Corner Ep: 59 Carin Hercock

    Join host Mike Bayly and guest Carin Hercock, Country Manager NZ at Ipsos, to explore what the June AI Monitor study of 32 countries and 1,000 New Zealanders reveals about a country that keeps learning more about AI and keeps getting less excited about it. You'll hear why 65% of Kiwis report nervousness against only 40% excitement, why just 21% trust AI outputs enough to skip double-checking them compared with a 31% global average, and why only 35% believe the benefits to society outweigh the environmental cost, one of the lowest results measured anywhere. Plus her view on the Anglosphere cluster that puts New Zealand alongside Canada, Britain and Australia and far from the more enthusiastic markets in Asia and Latin America, the data sovereignty trade-off nobody has connected to the data centre debate, and why government and employers now carry the job of framing AI as something other than a payroll reduction exercise. In this episode we discuss: • 00:00 – Intro & guest bio• 00:45 – How the AI Monitor is run: 32 countries, 1,000 Kiwis, 18 plus• 01:30 – 65% nervous, 40% excited, 72% claim they understand AI• 02:00 – "Sceptically optimistic": why understanding rises as excitement falls• 02:30 – Power used for good or ill, and the time spent checking outputs• 03:30 – Saving time at work but expecting a worse job market• 04:00 – Consultancies dropping graduates and government headcount messaging• 05:00 – The growth-opportunity case the headlines miss• 06:00 – Framing AI as a leadership failure rather than a technology problem• 06:53 – Reskilling tasks versus losing whole jobs• 07:30 – The Anglosphere cluster: NZ, Canada, US, Britain, Australia• 08:00 – Economic growth potential and ageing populations as explanations• 09:00 – Leapfrogging: why less legacy infrastructure means faster adoption• 10:00 – NZ plateaued while Canada, Britain and the US went backwards• 11:00 – Anthropic's Economic Index: NZ ranked fifth of 121 countries• 12:00 – Super users snowballing while others feel left behind• 13:00 – Only 21% trust AI enough to skip double-checking outputs• 14:00 – Testing and tricking the model, and AI plus human intelligence• 15:30 – Legacy distrust from earlier models, and knowledge decay worries• 16:30 – 35% say benefits outweigh environmental cost, against 49% globally• 17:30 – Data centres, energy draw, and protecting the tourism economy• 18:30 – The data sovereignty trade-off nobody has joined up• 19:30 – Only 39% trust companies to protect their data• 20:30 – Where Kiwis accept AI: research and restaurant orders versus hiring and health• 22:00 – Disclosure: 85% want AI use declared up front• 22:30 – The biggest surprise: three flat years against soaring AI valuations• 24:00 – "We trained the team and gave them the tools, but nothing changed"• 25:00 – Why individual training alone will not shift an organisation• 26:30 – Trench or trend: where sentiment goes from here• 28:00 – Where to find the report and get in touch• 28:45 – Outro & next steps CARIN'S LINKEDIN: https://nz.linkedin.com/in/carin-hercock-20118313 SUBSCRIBE TO NZ'S AI NEWS: https://theaicorner.co.nz/subscribe

  2. Jul 12

    How Halter supports its sales and customer support with AI agents | The AI Corner Ep: 58 Pouria Asghari

    Join host Mike Bayly and guest Pouria Asghari, Head of GTM Ecosystems & AI at Halter, to explore how a fast-scaling agritech unicorn builds an AI engine that runs its go-to-market from first support ticket to field sales visit. You'll hear how Halter triages farmer queries through a RAG pipeline over nearly 500 knowledge modules, why animal-welfare questions trigger instant escalation to a human, and how the sales team dictates voice notes in the ute that Whisper and Claude turn into clean CRM records. Plus his view on why customer experience beats cost so they run on Opus rather than Sonnet, why specialists beat rare full-stack unicorns, and why the real prize is killing the copy-paste admin tax rather than replacing people. In this episode we discuss: • 00:00 – Intro & guest bio• 00:45 – What "Head of GTM Ecosystems & AI" actually means• 02:30 – Was the function built from day one, or a response to scale?• 04:00 – The breaking point: support tickets 3-4x as onboarding grew• 05:30 – Deciding what to automate versus keep human on animal welfare• 06:53 – Testing internally, then piloting with four or five farmers• 08:00 – The 96% CSAT bar and low threshold for escalation• 08:37 – Behind the scenes: how the support agent is built• 09:20 – "How do I draw my first virtual fence?" walked through end to end• 10:00 – The decision tree: knowledge, escalation, data, or sales query• 11:17 – When cattle break the boundary: instant escalation• 12:21 – Contextualising answers by where the farmer is in onboarding• 13:44 – Two-way clarification when the RAG match score drops below 70%• 14:25 – Why they chose RAG, and rethinking the pipeline as they scale• 15:23 – Model strategy: an Anthropic house, Whisper for voice• 16:25 – Fable for planning and code, and the week and a half it went dark• 17:27 – Managing cost: over-indexing on quality, no routing to Sonnet• 19:00 – Field sales voice notes turned into CRM records in the ute• 20:03 – Meeting recordings with speaker separation for farmer context• 21:03 – Learning loops: feeding ticket data back to product and training• 24:25 – Team shape: product, engineering, and a dedicated content arm• 26:17 – Building GTM AI teams: MarTech and Salesforce roles still matter• 28:43 – Specialists over unicorns, and why data must come before automation• 31:21 – Why you buy the CRM and build the intelligence layer on top• 32:15 – The vision: reducing admin so teams get hours back• 34:20 – Key takeaways: the copy-paste tax and quality-first models• 35:47 – How to reach Pouria and Halter's open roles• 36:19 – Outro & next steps POURIA'S LINKEDIN: https://nz.linkedin.com/in/pouria-asghari SUBSCRIBE TO NZ'S AI NEWS: https://theaicorner.co.nz/subscribe

  3. May 10

    Justin Flitter reveals why NZ businesses keep stalling on AI, the literacy and confidence gap, and how to shift from ChatGPT-as-Google to AI in the loop | The AI Corner Ep: 54 Justin Flitter

    Join host Mike Bayly and returning guest Justin Flitter, founder and Chief AI Officer of NewZealand.AI, to unpack why most Kiwi businesses are still using AI as a glorified Google search and how to shift from scattered shadow adoption to trusted, scalable systems. You'll hear Justin's view on the literacy and confidence gap, how an engineering firm cut a three-hour reporting task to ninety seconds with a Copilot agent, plus where IT ownership of AI should sit, the case for AI-in-a-box tools, and a preview of his eight-city AI for Business Masterclass roadshow. In this episode we discuss:• 00:00 – Intro & guest bio• 01:00 – AI adoption results across NZ• 03:30 – ChatGPT as Google replacement• 06:00 – Literacy versus confidence gap• 09:00 – Are we asking too much of people• 12:00 – AI for business in a box• 14:30 – Roadshow eight-city tour preview• 17:30 – Featured speakers and case studies• 20:00 – Where IT ownership should sit• 23:30 – Engineering firm proof point• 27:00 – Building momentum through proof points• 30:00 – Outro & next steps JUSTIN'S LINKEDIN: https://www.linkedin.com/in/justinflitter/ MIKE'S LINKEDIN: ⁠⁠https://nz.linkedin.com/in/mikebayly⁠⁠ SUBSCRIBE TO THE AI CORNER NEWSLETTER: ⁠⁠https://theaicorner.co.nz/subscribe

  4. Apr 27

    Node co-founder Steven Baker on diagnosing broken workflows, the disappearing CIO role, and building an AI-native company from New Zealand | The AI Corner Ep: 53 Steven Baker

    Join host Mike Bayly and guest Steven Baker, co-founder of Node (formerly Vendor Sage), to explore how he's building an AI-native company that acts as a CIO in your pocket. You'll hear why the CIO role is disappearing in mid-market, how Node's "curiosity loop" diagnoses workflows by connecting to Notion, Slack, CRMs, and HRIS systems, plus Steven's take on why 70% of software implementations fail (and 90% of AI ones), why shadow IT isn't the real problem, and how he's building a transparent, bootstrapped business with customers like Halter, Tracksuit, and Partly. In this episode we discuss:• 00:00 – Intro & guest bio• 01:30 – From Salesforce to Vendor Sage to Node• 03:00 – What Node actually does• 06:00 – The curiosity loop and discovery• 09:00 – Mapping business goals to workflows• 13:00 – Data privacy and trust concerns• 18:00 – Avoiding regression to the mean• 21:00 – Customer base and growth• 25:00 – Shadow IT, AI sprawl, and source of truth• 30:00 – Multi-agent architecture under the hood• 33:00 – Pricing strategy and bootstrapped growth• 38:00 – Outro & next steps STEVEN'S LINKEDIN: https://www.linkedin.com/in/steven-baker-395855128/ MIKE'S LINKEDIN: ⁠https://nz.linkedin.com/in/mikebayly⁠ SUBSCRIBE TO THE AI CORNER NEWSLETTER: ⁠https://theaicorner.co.nz/subscribe

  5. Apr 19

    AI, copyright, and why developers can't give an IP indemnity on AI‑built product anymore | The AI Corner Ep: 52 Andrew Dentice

    Join host Mike Bayly and guest Andrew Dentice, Partner at Hudson Gavin Martin, for a grounded look at the legal reality behind AI deployment in New Zealand. Andrew unpacks what AI procurement, governance and IP actually look like day-to-day, why developers building with generative models struggle to give clients a standard IP indemnity, and where agentic tools sit in the risk picture. You'll hear his take on the EU AI Act for NZ exporters, the shift from intelligence work to judgment work inside law firms, and why AI is quietly widening access to legal advice. • 00:00 – Intro & guest bio• 01:20 – What Hudson Gavin Martin actually does• 03:20 – Procuring Copilot vs bespoke AI builds• 05:30 – Why AI IP risk is different• 08:00 – Who's liable when AI infringes copyright• 11:30 – Output risk and agentic oversight• 12:45 – Indemnity gaps in AI builds• 15:00 – NZ as a regulation price-taker• 17:50 – EU AI Act and NZ exporters• 21:00 – Intelligence work vs judgment work• 27:30 – Juniors, Harvey and AI-native lawyering• 31:30 – Future of the entry-level lawyer• 37:40 – AI widening access to legal advice• 41:30 – Outro & next steps ANDREW'S LINKEDIN: https://www.linkedin.com/in/andrew-dentice/MIKE'S LINKEDIN: https://nz.linkedin.com/in/mikebaylySUBSCRIBE TO THE AI CORNER NEWSLETTER: https://theaicorner.co.nz/subscribe

About

Spotlighting Kiwis moving the local and global AI ecosystem forward.