Kia ora! Welcome to New Zealand’s weekly roundup of AI news, events, jobs and education.
Most AI tool videos and walkthroughs are polished.
But that’s not how real work gets done.
Introducing to the 30-min AI Challenge:
→ Each episode starts with a random task generated by AI.
→ I can use only AI tools to complete it.
→ No prep. No edits. All in 30-mins.
→ I’ll share the uncut footage: every tab-switch, glitch, and tool swap.
Challenge #1: Design a social media campaign for a vitamin C skincare product launch.
Tools used:
ChatGPT: Thinking and using ‘Projects’ for storing context
Suno: Creates music and jingles
Gamma: Builds decks and docs
Lovable: Designs visuals & apps fast
Perplexity Comet: Researches and verifies info
Reve: Image generation at scale
Gemini Nano Banana: Generates branded images
Gemini Veo 3.1: Produces short videos
Wispr Flow: Voice dictation tool
These experiments aren’t about polish, they’re about showing how fast you can move from nothing to something with the right stack.
This week’s highlights:
Leo Garcia Curtis: AI voice agents
The AI Agent hype is out of control
Weekly Tip: How to use Claude Skills
Health NZ rolls out AI scribe to 1,000 staff
RNZ uses AI to recreate dead man's voice
Mindhive's leather grading AI gets Series A
Iwi leaders launch AI bias safety framework
Huckleberry lands $2.1M for voice feedback
AI tools fuel 416% spike in NZ web skimming
AI adoption focus: redesign roles, not just headcount
Happy reading and listening ✌️
Did someone forward you this? Sign up!
HEADLINE STORY
⚠️ Only 20% of NZ CEOs are making AI a top strategic priority
📣 Word On The Street: AI is the top priority for NZ leaders investing in resilience, with only 20% of CEOs now leading the strategy personally.
🔎 Zooming In:
Despite the low CEO led AI direction, 62% of NZ firms are investing heavily in AI governance and 58% in AI infrastructure.
AI's biggest impact today is in IT and cybersecurity.
In three years, AI is set to revolutionise product development and R&D.
🏘️ Our Take: The data reveals that New Zealand’s CEOs are approaching AI with a prudent approach, but this caution currently risks limiting their future growth. The intent is there from CEOs aiming to lift the priority of new products and R&D from 16% to an aggressive 46% in three years. This target rightly acknowledges AI as the essential accelerator, the path to efficient, rapid R&D.
Currently, investment is heavily weighted toward establishing a responsible foundation for AI. The focus is on the necessary defensive work: 62% investing in governance and 58% in infrastructure. This high commitment to compliance and security is a valuable, risk-mitigating first step. It ensures that when AI is fully scaled, it will operate within a trustworthy, defensible framework.
The core challenge lies in accelerating the transition from this secure base to the intended growth outcome. With only 20% of companies reporting a CEO-led strategy, the executive mandate for an offensive AI plan is lagging behind the defensive investment. We are seeing a practical sequence of events where AI is being used for IT and cybersecurity first. However, the anticipated three-year wait to transition AI into the front-room function of product creation and growth creates an opportunity gap.
The strategic task now is not to scrap the current prudent investment, but to reclassify it as the launchpad. CEOs need to quickly convert their robust defensive foundation into an engine for competitive advantage. By establishing an immediate, bolder AI mandate from the top, they can compress the timeframe, allowing businesses to build their instinctive innovation muscle now, rather than waiting for global competitors to set the final technology standard. The goal is to move beyond simply managing AI risk to actively leveraging AI readiness for outsized commercial gain.
💼 Business & Industry
AI tools cause 416% spike in NZ web skimming attacks. Gen’s Q3 2025 report shows AI-built "VibeScams" fuelled an 82% global rise in breaches, with NZ web skimming attempts surging 416%. Attackers inject code to steal card details during checkout.
4-min read.
Our take: The paradox is that AI is simultaneously the primary weapon used by cybercriminals to automate highly convincing scams and the only viable shield for defenders to detect and neutralise those same threats.
Kiwi AI voice feedback tech (Huckleberry) secures $2.1M. Huckleberry raised its pre-seed, led by Oregon Venture Fund, for its voice-based 360 platform. Huckleberry creates an AI voice-based platform for continuous work feedback.
3-min read.
Our take: Voice-based HR tech makes continuous feedback a low-friction habit. Embedding AI into the feedback loop creates a continuous growth mindset at scale, enabling faster feedback loops among teams.
NZ AI leather grading AI (Mindhive) secures Series A funding. Mindhive Global, an agritech firm, closed its Series A, led by Cultivate Ventures, to expand AI-driven leather quality assessment. It already partnered with JBS Couros, the world's largest leather processor. Mindhive Global develops AI systems for precision leather hide grading.
1-min read.
Our take: Nothing gets us more excited than seeing traditional industries evolve with technology. This investment validates the pivot from manual grading to verified data intelligence. Scaling this tech across Brazil and Europe positions Mindhive as the global quality standard.
⚖️ Government & Legal
Health NZ rolls out AI scribe to 1,000 emergency staff. Licenses procured for frontline emergency doctors and nurses nationwide. Hawke's Bay and Whanganui pilots showed doctors saw one extra patient per shift.
2-min read.
Our take: Healthcare productivity gains are measurable and immediate. Expect demand for AI clinical documentation tools to surge as other specialties watch emergency departments win back time.
🔍 Education & Society
RNZ recreates dead man's voice with AI. The broadcaster used AI to let convicted murderer Ross Appelgren plead innocence posthumously in new podcast Nark. Only his written words were used. His widow approved, saying it sounds "just like him."
5-min read.
Our take: We're entering territory where consent from the deceased's family determines whether AI can speak for the dead. That's a significant shift in how we handle legacy. Media organisations will face commercial pressure to use AI voices instead of actors. This could set an interesting precedent.
Iwi leaders launch AI bias safety framework. The new framework ensures AI systems do not entrench bias or colonial extraction. It demands "free prior consent" for all personal data used in AI development.
3-min read.
Our take: This framework will shape the entire NZ data ecosystem. Companies using Crown data will need full transparency to prevent bias and uphold data as a taonga. Embedding Māori leadership and data sovereignty is the path to world-class ethical AI. This is a unique opportunity to build a trustworthy and responsible digital future.
📚️ AI for Business
Helping leaders and teams adapt, learn, and scale with AI.
1️⃣ Will Chinese LLMs commoditise the AI layer?: When major firms choose open-weight models for core services over proprietary APIs, the trust gap could become a margin gap. This shift suggests AI is rapidly becoming a cost-of-operations advantage.
1-min read.
2️⃣ Sharing to AI rather than socials: Brands are now optimising for model memory, not human clicks. They're injecting positioning straight into ChatGPT and Claude, creating “source of truth” pages for LLMs, trying to hijack how models describe them. The question now is how platforms will police prompt injection.
1-min read.
3️⃣ One NZ CEO “AI is coming to your job, not for your job”: Jason Paris nailed it. We’re not in the “AI agent takes your role” moment. We’re in the “AI removes the pointless parts of your role” moment. The risk isn’t mass layoffs, it’s leaders who don’t redesign roles, don’t map AI to strategy, and don’t build a learning culture.
1-min read.
4️⃣ The AI Agent Hype is Out of Control. Most new AI Agent tech is hype: 2025 was sold as “the year of AI agents”. Reality: most “agents” are glorified workflows in trench coats. In testing, the best agent only automated ~2.5% of freelance tasks. Leaders should focus on workflow automation & capability design, not fantasy headcount cuts.
4-min read.
5️⃣ Conviction in AI comes from practice, not persuasion: To shift the reluctant observer persona, avoid evangelism and mandate two hours weekly of structured AI practice. Genuine intuition forms when experience replaces intellectual argument.
1-min read.
6️⃣ AI Innovation happens in the margins, not R&D: Curious employees are testing tools faster than formal training allows, making grassroots experimentation your real AI lab. Create safe sandboxes, or risk bureaucracy masquerading as governance.
84-sec watch.
🎙️ The AI Corner Podcast
This week's guest is Leo Garcia Curtis, Founder of Maximize AI Agency. Hear:
Why Leo left his agency to go all‑in on AI.
Consulting vs building: the two phases of AI adoption.
Why New Zealand's small market is actually an advantage.
🌍 Tech Updates From Global
The selected top headlines from each major AI tech company.
OpenAI
Reorganised into a for-profit Public Benefit Corporation ("OpenAI Group"), enabling capital expansion while maintaining non-profit governance through the OpenAI Foundation.
SoftBank invested $22.5B into OpenAI, contingent on IPO and structural reform; Microsoft retains 27% stake (~$135B valuation) with control provisions on AGI.
Developing a generative music tool in collaboration with Juilliard students to turn text/audio into music, competing with tools like Suno.
Released Aardvark, a GPT-5-powered autonomous AI agent to detect and fix software vulnerabilities, now in private beta.
Offering free ChatGPT Go subscriptions to all users in India to accelerate international user growth.
Reported that ~0.07% of its 800M weekly users show signs of psychosis or mania, prompting new global safety efforts.
To roll out Gemini Visual Layout with modular, interactive cards for productivity apps.
Released Pomelli, an AI tool for generating marketing campaigns based on brand website content.
Added new features to NotebookLM, including larger context window, customisable personas and higher response quality.
Released a free AI curriculum in collaboration with DeepMind and UCL.
Reported 34% year-on-year cloud revenue growth, driven heavily by AI adoption in enterprise settings.
Anthropic
Introduced finance-specific Agent Skills that can explain spreadsheets, fix formulae, and build new workbooks from scratch.
Opened first Asia-Pacific office in Tokyo; signed agreement with Japan AI Safety Institute.
Microsoft
Incorporating AI features into Outlook for major user experience revamp under new leadership.
Introduced Researcher tool to Microsoft 365 Copilot, enabling users to browse, code, and analyse data with a virtual machine.
Rolled out new Azure tools to help office workers build business apps via low-code/no-code interfaces.
Facing legal action from Australia over AI-related customer service transparency issues.
Implemented layoffs affecting AI-adjacent corporate roles as part of restructuring.
Meta
Introduced new AI tools for editing Instagram Stories (photos and videos) directly in-app.
Plans to raise up to $30B through bond offerings to finance continued AI infrastructure investments.
Deployed AI automation for compliance review processes to improve internal efficiency.
Amazon
Launched Creative Studio, an AI ad-generation tool for mid-market brands across Amazon platforms including Prime Video.
Announced plans to invest $11B in a new AI-focused data centre in Indiana.
Cut approximately 14,000 corporate roles, heavily impacting AI and AWS teams in a broader operational restructuring.
xAI
Released Grokipedia v0.1, an AI-written encyclopaedia with over 900,000 articles and live-edit functionality.
NVIDIA
Became the first public company to reach a $5 trillion market valuation—16% of US GDP, driven by AI chip dominance.
Forecasted $500B in upcoming revenue from new generations of chips.
💡 Tip of the Week: Claude Skills turn your workflows into reusable AI systems
Forget one-off prompts. Claude Skills let you package entire workflows, complete with instructions, code, and assets, into a single reusable AI capability.
Why it matters
→ Prompts are disposable. Skills are permanent.
→ Each Skill loads automatically when relevant, no re-explaining context.
→ You can bundle logic, frameworks, and brand rules for consistent output.
→ They compound. One Skill feeds another, building your own AI stack.
How to do it
Go to Settings → Capabilities → Skills in Claude.
Use the Skill Creator (it’s a meta-skill) to generate your first Skill folder, or create your own.
Add a
SKILL.mdfile defining what the Skill does and when Claude should use it.Bundle any reference docs, scripts, or assets you need.
Upload, activate, and test. Claude now runs that workflow anywhere.
Think of Claude Skills as your personal automation library, each one a trained assistant that knows how you work and executes with precision every time.
🕒 More details here.
📅 AI Events in New Zealand
A light event week, with only 10 across the country.
This week’s featured event:
🔦Spotlight: Autohive Intermediate AI Agent Cooking Class, Friday, Wellington: A hands-on workshop for AI practitioners ready to graduate from single GPTs to orchestrated multi-agent workflows.
📅 Promote your event with us. Reply to let us know.
💼 AI Roles Around Aotearoa
Picklist of 🌶️ HOT 🌶️ new roles in AI this week.
AI Product Manager, BitMart: Remote
AI Solution Specialist, PWC: Wellington
AI Senior Consultant, Deloitte: Christchurch
💼 Promote your job with us. Reply to let us know.
🤦 ️ AI Fail Of The Week
We all love AI, but it’s certainly far from perfect 🤔 …
👋 Mike & Erin