EP 15: AGI vs Agentic AI — What Actually Matters for Business in 2026

EP 21: AI in Your Day - The Unglamorous Workflows That Actually Save Hours

How many times did you actually use AI yesterday, and did any of it save you real time? In Episode 21 of ⁠Zero Shot⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ [https://zeroshot.com.au/] , ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠James Gauci⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ [https://www.linkedin.com/in/jamesrgauci/] and ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Llew Jury⁠ [https://www.linkedin.com/in/llewjury/] pull back the curtain on their own daily AI workflows: the boring, unglamorous, hour-saving habits that rarely make it into the headlines. From doing email entirely by voice to automated morning client briefings to turning accounting software into an on-demand CFO, this episode is a practical look at what AI in a real working day actually looks like in 2026. Recorded with a live update from the Super AI conference in Singapore, it closes on the one high-value AI application almost nobody in a corporate role is using yet. Key Highlights * 75% of knowledge workers use AI, but only 36% save real time * Doing email by voice can turn two hours into twenty minutes * Automated morning briefings handle client prep before 6 am * Connecting Claude Code to accounting software produces CFO grade financial reports * Microsoft Copilot's Frontier agent setting unlocks Anthropic models inside the stack * Connectors and skills are the biggest unlock for busy Claude users * The shift from T-shaped to square-shaped generalists is already here * Knowing what good looks like is the skill AI cannot replace Tools and Frameworks Mentioned * Granola: [https://www.granola.ai⁠] AI meeting notes tool that records audio locally and feeds notes into other AI workflows for summaries and content. * Wispr Flow [https://wisprflow.ai⁠]: Voice dictation tool that runs your speech through an LLM to clean up grammar and misheard words before sending to your AI of choice. * Microsoft Copilot Cowork [https://www.microsoft.com/en-us/microsoft-365-copilot⁠]: The agentic layer inside Microsoft Copilot, activated via the Frontier program, that orchestrates across Outlook, SharePoint, and OneDrive. * Claude Connectors and Skills: Claude's customisation features for hooking up business systems like 365, Gmail, CRMs, and QuickBooks, and building a reusable skills library. * Model Context Protocol (MCP) [https://modelcontextprotocol.io⁠]: The open standard that lets tools like Claude Code connect deeply to systems such as Xero via CLI, accessing far more than standard connectors allow. Connect with the Hosts: * ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠James Gauci:⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ [https://www.linkedin.com/in/jamesrgauci/] CEO of Cadent. * ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Llew Jury:⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ [https://www.linkedin.com/in/llewjury/] Managing Director of Advancer at the AI Agency. * ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Zero Shot:⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ [https://zeroshot.com.au/] Follow us on LinkedIn for clarity in a complex landscape. 🤠 Sponsor: This episode is brought to you by ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Prompt Cowboy⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ [https://www.promptcowboy.ai/], the agentic prompting tool that helps teams build structured, precise prompts before they hit whichever model they are using. Head to promptcowboy.ai and stop leaving results on the table. Produced by ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Yennia La Rotta⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ [https://www.linkedin.com/in/zarich-la-rotta/].⚡

EP 20: How AI Memory Actually Works - Context, Data Sovereignty, and What Your AI Is Remembering

What is your AI actually storing about you, and where does that information live? In Episode 20 of Zero Shot⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ [https://zeroshot.com.au/] , ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠James Gauci⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ [https://www.linkedin.com/in/jamesrgauci/] and ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Llew Jury [https://www.linkedin.com/in/llewjury/] go under the hood of one of the most misunderstood features in AI today: memory. From why AI systems seem to forget things mid-conversation to the difference between context windows, persistent memory, and projects, this episode breaks down how memory actually works across the major platforms and what the implications are for businesses handling sensitive data. This is a practical, no-jargon guide to understanding and managing the memory layer of your AI tools. Key Highlights * Context windows determine how much an AI reliably remembers per session * Google was first to release a million token context window, changing agentic tasks * Persistent memory is a text file updated by the model, not human-like recall * Data stored in US-based AI tools can be accessed by US law enforcement without a warrant * James keeps memory turned off in Claude for deliberate security reasons * One project per context area eliminates most mixing and confusion problems * Memory portability between platforms is becoming a competitive differentiator * Context engineering is emerging as a new field built on decades of data governance Tools and Frameworks Mentioned Claude Projects [https://claude.ai⁠] - A persistent context feature inside Claude that lets users build dedicated memory, files, and instructions per topic or client. Available inside the Claude app and web interface. Beelink Mini PCs [https://www.bee-link.com⁠] - Chinese mini PC manufacturer producing local AI hardware for running open source models offline, referenced as part of the emerging consumer agentic hardware culture. Micro AGI [https://microagi.de⁠] - German startup offering a free home cleaning service funded by harvesting in-home behavioral data via head-mounted cameras for frontier model training. Referenced as an example of extreme data collection tradeoffs. GDPR (General Data Protection Regulation) [https://gdpr.eu⁠] - EU regulation giving individuals the legal right to request deletion of their personal data from any company operating in Europe, including major AI frontier labs. Connect with the Hosts: * ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠James Gauci:⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ [https://www.linkedin.com/in/jamesrgauci/] CEO of Cadent. * ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Llew Jury:⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ [https://www.linkedin.com/in/llewjury/] Managing Director of Advancer at the AI Agency. * ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Zero Shot:⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ [https://zeroshot.com.au/] Follow us on LinkedIn for clarity in a complex landscape. 🤠 Sponsor: This episode is brought to you by ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Prompt Cowboy⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ [https://www.promptcowboy.ai/], the agentic prompting tool that helps teams build structured, precise prompts before they hit whichever model they are using. Head to promptcowboy.ai and stop leaving results on the table. Produced by ⁠⁠⁠⁠⁠⁠⁠⁠⁠Yennia La Rotta⁠⁠⁠⁠⁠⁠⁠⁠⁠ [https://www.linkedin.com/in/zarich-la-rotta/].⚡

EP 19: ChatGPT, Claude, Gemini or Grok? The Honest Answer for Business

Which AI model is actually worth your time and money in 2026, and how do you decide? In Episode 19 of Zero Shot⁠⁠⁠⁠⁠⁠⁠⁠⁠ [https://zeroshot.com.au/] , ⁠⁠⁠⁠⁠⁠⁠⁠⁠James Gauci⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ [https://www.linkedin.com/in/jamesrgauci/] and ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Llew Jury ⁠ [https://www.linkedin.com/in/llewjury/] do the episode the audience has been asking for: a real, no-benchmarks, no-lab-scores breakdown of the four major frontier AI models. This is not a comparison of leaderboard scores or press releases. It is a practical, values-informed assessment of ChatGPT, Claude, Gemini, and Grok from two people who have all four open on their desktops every day and use them across client work, internal builds, and AI training. The goal is simple: help you make a smarter model decision for your business. Key Highlights * ChatGPT built the market but governance controversies have changed perceptions * Anthropic's run rate now exceeds OpenAI's at thirty billion US annually * Google invented the transformer architecture and remains a far from dark horse * Grok has the fewest guardrails and the most governance risk for Australian businesses * The model is rarely the differentiator; the prompt and the ecosystem are * AI tourism, jumping between models without committing, is a real productivity drain * Your conversation data is stored for thirty days and accessible without a warrant in the US * Choosing a model is a vendor decision, not just a tool decision Tools and Frameworks Mentioned * Granola [https://www.granola.ai⁠] - AI meeting notes tool that records audio locally on your device rather than sending a visible bot into the meeting. Raised a $125 million Series C. * AWS Bedrock Nova Models [https://aws.amazon.com/bedrock⁠] - Amazon's vertically integrated foundation models, referenced for strong agentic workflow performance and on-shore Australian inference options for regulated industries. * LLM Council [https://llmcouncil.com⁠] - A lightweight tool that lets you query multiple AI models simultaneously via plugged-in APIs and compare responses side by side. * Gemma 4 (Google DeepMind) [https://deepmind.google/technologies/gemma⁠] - Google's open source model series, noted as the best US open source model currently available and usable on consumer hardware. * Kimi K2 (Moonshot AI) [https://www.moonshot.ai⁠] - A low-cost thinking model flagged as a practical alternative for high-volume, repetitive workflow tasks where frontier model power is unnecessary. Connect with the Hosts: * ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠James Gauci:⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ [https://www.linkedin.com/in/jamesrgauci/] CEO of Cadent. * ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Llew Jury:⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ [https://www.linkedin.com/in/llewjury/] Managing Director of Advancer at the AI Agency. * ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Zero Shot:⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ [https://zeroshot.com.au/] Follow us on LinkedIn for clarity in a complex landscape. 🤠 Sponsor: This episode is brought to you by ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Prompt Cowboy⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ [https://www.promptcowboy.ai/], the agentic prompting tool that helps teams build structured, precise prompts before they hit whichever model they are using. Head to promptcowboy.ai and stop leaving results on the table. Produced by ⁠⁠⁠⁠⁠⁠⁠⁠Yennia La Rotta⁠⁠⁠⁠⁠⁠⁠⁠ [https://www.linkedin.com/in/zarich-la-rotta/].⚡

EP 18: AI Inaction - The Real Cost to Your Business and How to Recover

What is it actually costing an organisation to keep having conversations about AI without acting on them? In Episode 18 of Zero Shot⁠⁠⁠⁠⁠⁠⁠⁠ [https://zeroshot.com.au/] , ⁠⁠⁠⁠⁠⁠⁠⁠James Gauci⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ [https://www.linkedin.com/in/jamesrgauci/] and ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Llew Jury [https://www.linkedin.com/in/llewjury/] go after one of the most uncomfortable topics in business right now: the anatomy of AI inaction. From Japan losing the equivalent of $78 billion a year due to stalled modernisation to Australian organisations repeating the same cybersecurity mistakes that defined the last decade, this episode maps out what paralysis actually costs at the business level and what organisations successfully closing the gap are doing differently. No benchmarks, no lab scores, just real patterns from the front line of AI adoption. Key Highlights * Japan lost $78 billion annually from delayed technology modernisation * Analysis paralysis in the boardroom kills AI momentum faster than budget constraints * Accountability is the first governance control most organisations skip * Copilot training in isolation does not constitute an AI strategy * Entry level technology hiring is dropping while AI native graduates go ignored * The best AI talent is already leaving organisations perceived as standing still * Successful recovery starts with one use case, one metric, one owner * The compound return on early AI implementation accelerates faster than most expect Tools and Frameworks Mentioned * Hermes [https://github.com]— A lightweight open source computer use agent, positioned as a more manageable and secure alternative to OpenClaude for organisations considering agentic AI deployments. Available via GitHub at https://github.com * Claude Code [https://claude.ai/code] — Anthropic's terminal based agentic coding tool, used to build and deploy internal platforms directly from the command line. * ISO 27001 [https://www.iso.org/standard/27001]— The international standard for information security management systems, referenced as a benchmark for organisations building credible security and AI governance foundations. * Notifiable Data Breach Scheme [https://www.oaic.gov.au/privacy/notifiable-data-breaches]— Australia's mandatory breach reporting framework under the Privacy Act, used as a case study for the real regulatory consequences of cybersecurity inaction. Connect with the Hosts: * ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠James Gauci:⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ [https://www.linkedin.com/in/jamesrgauci/] CEO of Cadent. * ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Llew Jury:⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ [https://www.linkedin.com/in/llewjury/] Managing Director of Advancer at the AI Agency. * ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Zero Shot:⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ [https://zeroshot.com.au/] Follow us on LinkedIn for clarity in a complex landscape. 🤠 Sponsor: This episode is brought to you by ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Prompt Cowboy⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ [https://www.promptcowboy.ai/]—the fastest way to turn everyday business into automated workflows. They wrangle the hard stuff so you can focus on what matters. Visit promptcowboy.ai to start saving time today. Produced by ⁠⁠⁠⁠⁠⁠⁠Yennia La Rotta⁠⁠⁠⁠⁠⁠⁠ [https://www.linkedin.com/in/zarich-la-rotta/].⚡

EP 17: Who's Behind Zero Shot - The Founders, the Failures & the AI Journey

What does it actually look like to build an AI business from the inside out? In Episode 17 of Zero Shot⁠⁠⁠⁠⁠⁠⁠ [https://zeroshot.com.au/] , ⁠⁠⁠⁠⁠⁠⁠James Gauci⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ [https://www.linkedin.com/in/jamesrgauci/] and ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Llew Jury [https://www.linkedin.com/in/llewjury/] do something they have never done before on this show: they turn the lens on themselves. After 16 episodes covering the tools, the models, the governance debates, and the industry shifts, this is the honest, unscripted account of where both businesses actually came from, what went wrong, what the dayyad-ot- of running an AI consultancy looks like in 2026, and what keeps two founders going when the market does not cooperate. No client stories borrowed for credibility. Just the real version. Key Highlights * James went from performing arts and psychology to AI governance * Llew built and sold two agencies before founding Advancer * Cadent bet big on AI regulation and a ministerial change ended it * IBM Watson cost Llew and his team nearly 50K and barely worked * Fail fast, fail quietly, fail cheaply, fail internally is still the principle * Process mapping from 20 years ago remains the foundation for AI consulting today * The first question to any AI brief is what are you trying to achieve * Proof of value, not proof of concept, is how serious AI adoption starts Tools and Frameworks Mentioned * Advancer [https://advancer.com.au]- Llew Jury's AI agency specialising in workforce capability, AI training, and business transformation. They work directly with leadership teams to cut through the hype and deliver measurable AI adoption outcomes. * Cadent [https://cadent.au] - James Gauci's AI consultancy focused on governance, security, and engineering AI solutions for complex, high stakes industries. From policy design to technical implementation, they help organisations adopt AI responsibly. * ISO 42001 [https://www.iso.org/standard/81230.html]- The international standard for AI management systems, which Cadent built internal tooling around and still uses as a governance framework. * Ethy - Cadent's internal AI governance and compliance tool, built as an ISO 42001 aligned product before the Australian regulatory window closed. Not publicly available. * Swimlane Process Mapping - A whiteboard based business analysis method for mapping actors, processes, and outputs, referenced by Llew as the foundation for every AI engagement. No single link; universally documented across business analysis and management consulting literature. Connect with the Hosts: * ⁠⁠⁠⁠⁠⁠⁠⁠⁠James Gauci:⁠⁠⁠⁠⁠⁠⁠⁠⁠ [https://www.linkedin.com/in/jamesrgauci/] CEO of Cadent. * ⁠⁠⁠⁠⁠⁠⁠⁠⁠Llew Jury:⁠⁠⁠⁠⁠⁠⁠⁠⁠ [https://www.linkedin.com/in/llewjury/] Managing Director of Advancer at the AI Agency. * ⁠⁠⁠⁠⁠⁠⁠⁠⁠Zero Shot:⁠⁠⁠⁠⁠⁠⁠⁠⁠ [https://zeroshot.com.au/] Follow us on LinkedIn for clarity in a complex landscape. 🤠 Sponsor: This episode is brought to you by ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Prompt Cowboy⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ [https://www.promptcowboy.ai/]—the fastest way to turn everyday business into automated workflows. They wrangle the hard stuff so you can focus on what matters. Visit promptcowboy.ai to start saving time today. Produced by ⁠⁠⁠⁠⁠⁠Yennia La Rotta⁠⁠⁠⁠⁠⁠ [https://www.linkedin.com/in/zarich-la-rotta/].⚡