For twenty years, being good at your job meant learning how to navigate software. So what happens when AI is the one clicking the buttons?

In Episode 31 of Zero Shot⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ , ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠James Gauci⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ and ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Llew Jury unpack one of the biggest shifts coming for enterprise software: when AI becomes the primary user, the whole relationship between humans, apps, and the companies that sell them starts to change. If you no longer open the CRM, the finance platform, or the payroll engine yourself, and an AI moves across those systems for you, then the questions get uncomfortable fast. Why do we still price software per human seat? What are we actually paying software companies for? And where does the durable value move to when the interface matters less?

Key Highlights

• AI is becoming the intermediary between you and your data
• Software goes less visible, not less important, especially under compliance
• Per seat pricing breaks the moment AI does the work
• The shift from per user to per user plus consumption
• Vendors licensing models at roughly a fifty percent cut reshapes the economics
• Don't rip out software at renewal, know what the vendor actually owns
• Data, business logic, and integrations are the hard things to replace
• There will be losers, and there will be trillion dollar winners

Five Questions to Ask Before Your Next Software Decision

1. What are we actually paying for, access to a service or real work done?
2. How much of this still needs a human to interact with the app directly?
3. If AI performed more of this workflow, would per seat pricing still make sense?
4. What does this vendor actually own, your data, workflows, and business logic?
5. How is this vendor preparing for AI users, AI agents, and new pricing models?

Tools and Frameworks Mentioned

• Minimax on AWS Bedrock - A high-performing open weight model now available on Amazon's cloud, offering frontier performance without model-level vendor lock-in.
• Microsoft Copilot Cowork - The agentic, consumption-based layer inside Copilot, illustrating the move from per seat to per user plus consumption pricing.
• ChatGPT with Astra - OpenAI's flagship model, referenced alongside Sam Altman's comments on pacing releases around alignment and safety.
• AI Orchestration - Coordinates actions across multiple systems and workflows through AI.
• Human Up the Stack - Moves people from operating software to supervising judgment and outcomes.
• Performance-Based Software Pricing - Links software cost to usage, outcomes or completed work instead of seats.
• Software Stack Audit - Reviews what a business is actually paying for, what still requires human interaction and where AI could change the economics.
• Technology Investment Portfolio - Treats software, AI and internal development as different investments with different risk and return profiles.

Connect with the Hosts:

⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠James Gauci:⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ CEO of Cadent.

⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Llew Jury:⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ Managing Director of Advancer at the AI Agency.

⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Zero Shot:⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ Follow us on LinkedIn for clarity in a complex landscape.

🤠 Sponsor:

This episode is brought to you by Advancer, the AI agency that cuts through the hype to deliver real, measurable results. They help Australian businesses identify where AI can genuinely create value, then design and build the agents, automations, and custom AI solutions that make those workflows work in practice. With over a hundred clients across Australia, visit advancer.com.au to move from ideas to implementation.

Produced by ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Yennia La Rotta⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠.⚡

What if the biggest obstacle to AI isn’t the model, but the data underneath it?

In Episode 33 of ⁠⁠⁠⁠Zero Shot⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ , ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠James Gauci⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ and ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Llew Jury go beneath the AI interface to explore the data foundations that determine whether enterprise AI actually works. They unpack AI readiness, data quality, data governance, data lakehouses, structured and unstructured data, context windows and why the latest AI model cannot compensate for unreliable information underneath it.

Key Highlights

• Why AI projects increasingly fail because of data
• What good and bad data actually mean
• Why newer AI models cannot fix weak foundations
• How data lakehouses support modern AI systems
• The untapped value inside unstructured business data
• Why data governance is becoming a board issue
• How second screening validates AI generated business insights
• Why quality context matters more than data volume

Tools and Frameworks mentioned in this episode

• Databricks Lakehouse⁠ - Combines structured warehouse data with flexible unstructured data for AI workloads.
• Second Screening - Compares AI generated results against established reporting systems before increasing autonomy.
• Context Window Management - Limits AI to the most relevant information for more reliable, efficient outputs.
• Microsoft Fabric⁠ - Data and analytics platform supporting lakehouse architectures and enterprise AI workloads.
• Snowflake⁠ - Cloud data platform referenced for organising and accessing enterprise data for AI.

Connect with the Hosts:

⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠James Gauci:⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ CEO of Cadent.

⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Llew Jury:⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ Managing Director of Advancer at the AI Agency.

⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Zero Shot:⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ Follow us on LinkedIn for clarity in a complex landscape.

🤠 Sponsor:

This episode is brought to you by Cadent, the experts in embedding high ROI artificial intelligence into complex, high stakes industries. From technical development to responsible AI governance, they ensure your technology is perfectly aligned with your humanity. Visit cadent.au to build a more certain future for your organisation.

Produced by ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Yennia La Rotta⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠.⚡

What happens when AI gives us more capability, but less human agency?

Recorded live from the Green Room at Something Digital in Brisbane, James Gauci and Llew Jury explore three growing tensions in AI and business: automation versus human agency, bigger systems versus fragile trust, and greater intelligence versus better decision making. They unpack AI dependency, responsible AI adoption, workplace culture, governance and why community and authentic human connection may become more valuable as AI becomes more embedded in organisations.

Key Highlights

• When AI convenience starts becoming dependency
• Why human agency matters in AI adoption
• How automation can create unintended business bottlenecks
• Why bigger AI systems can create fragile trust
• The relationship between AI adoption and workplace culture
• Why more intelligence does not guarantee better decisions
• How AI governance can support responsible innovation
• Why community becomes more valuable in an AI world

Tools and Frameworks mentioned in this episode

Human Centred AI Design - Designing AI around affected people, stakeholders and real user needs.

AI Governance Frameworks - Creating shared definitions, policies and accountability for responsible AI adoption.

Stakeholder Impact Assessment - Evaluating how AI systems affect employees, customers and other stakeholders.

Explore Qwen⁠ - Alibaba’s open model family discussed as an alternative for running capable AI on your own infrastructure.

Explore ElevenLabs⁠ - Voice AI platform used in the Tivoli accessibility experience discussed in the episode.

Connect with the Hosts:

• ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠James Gauci:⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ CEO of Cadent.

• ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Llew Jury:⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ Managing Director of Advancer at the AI Agency.

• ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Zero Shot:⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ Follow us on LinkedIn for clarity in a complex landscape.

  
  

🤠 Sponsor:

This episode is brought to you by ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Prompt Cowboy⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠, 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.⚡

Standing with ColombiaWhat happens when AI uncertainty becomes a business risk?

In Episode 31 of ⁠⁠⁠Zero Shot⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ , ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠James Gauci⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ and ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Llew Jury discuss how AI anxiety ⁠is becoming part of the conversation around AI adoption, workplace transformation and leadership. In this episode, James Gauci and Llew Jury explore the behavioural science behind AI anxiety, its relationship with shadow AI, employee trust and resistance, and how organisations can approach AI change without losing human agency.

Key Highlights

• Why AI anxiety can be understood as business risk
• How workplace identity shapes reactions to AI adoption
• The connection between AI anxiety and shadow AI
• How unmanaged anxiety can influence AI adoption
• Why AI generated sameness can affect customer trust
• The growing competitive value of authentic human communication
• How transparency can strengthen confidence in AI assisted work
• The CAM framework for leading through AI uncertainty

Tools and Frameworks mentioned in this episode

CAM Framework - Communicate, Acknowledge and Maintain Agency when introducing AI change. Introduced in this episode.AI Readiness Assessment - Evaluates employee confidence, capability and concerns before and after AI adoption or training.

Claude Opus 5 - Anthropic model discussed for coding and agentic workflows.

ChatGPT - Generative AI platform referenced in workplace adoption and training.

Something Digital - Brisbane digital innovation festival referenced ahead of the next live episode.

Connect with the Hosts:

• ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠James Gauci:⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ CEO of Cadent.

• ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Llew Jury:⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ Managing Director of Advancer at the AI Agency.

• ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Zero Shot:⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ Follow us on LinkedIn for clarity in a complex landscape.

  
  

🤠 Sponsor:

This episode is brought to you by ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Prompt Cowboy⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠, 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.⚡

Every business is thousands of decisions stacked on top of each other. So what actually changes when AI pulls up a seat at the table?

In Episode 30 of ⁠⁠Zero Shot⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ , ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠James Gauci⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ and ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Llew Jury move past productivity and into something bigger: how AI is reshaping the way leaders make decisions. When people hear AI decision making, they assume it means AI makes the call, and that is the least interesting and most dangerous use of it. This episode is about using AI as a thinking partner to sharpen a decision, where it can quietly lead you astray through bias and confident errors, and why the decision and the accountability always stay firmly with you.

Key Highlights

• AI as a thinking partner, not the pilot of the decision
• Use it to compress the research phase, not replace your judgment
• Turn AI into an adversarial analyst to surface your blind spots
• Hiring done right: crunch the thinking, never outsource the verdict
• Automation bias, anchoring, and groupthink are the real traps
• Write your own view down before you let AI speak first
• Know your patch before trusting AI in a domain you don't own
• Accountability never transfers to a model, it stays with the human

Tools and Frameworks Mentioned

Meta Muse Glimmer: Meta's newly released open weight model, benchmarking close to the current Qwen generation and marking their return to open source AI.

RCTF Framework (Advancer): Llew's Role, Context, Task, Format prompt structure for briefing AI properly on complex decisions.

DMAIC and the Deming Cycle: Classic lean process management frameworks James applies to structured, higher-quality decision making with AI.

⁠⁠Shape the show⁠⁠

Connect with the Hosts:

• ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠James Gauci:⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ CEO of Cadent.

• ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Llew Jury:⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ Managing Director of Advancer at the AI Agency.

• ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Zero Shot:⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ Follow us on LinkedIn for clarity in a complex landscape.

  
  

🤠 Sponsor:

This episode is brought to you by ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Prompt Cowboy⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠, 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.⚡

You have probably nodded along in a meeting about autonomous AI agents without being totally sure what one actually is. You are not alone.

In Episode 29 of ⁠ ⁠⁠Zero Shot⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ , ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠James Gauci⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ and ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Llew Jury ⁠⁠ cut through the most hyped and least understood term in AI right now: the agent. This is the honest, no-jargon picture of what an AI agent really is, how you actually build one, where they earn their keep in a business, and where they quietly fall apart. From the difference between a chatbot, an automation, and a truly autonomous agent, to why roughly forty percent of agent projects get pulled, this episode gives you a clear way to decide what to hand an agent and what to keep a human on.

Key Highlights

• Autonomy is a dial, not a switch, and it defines what makes an agent
• The bright line: does it make a plan and use tools on its own
• Model plus context, prompt, tools, and memory equals an agent
• Chatbots and automations are often mislabelled as agents
• Most failed agent projects die from hype and poor scoping, not tech
• High performers gain from AI, low performers can go backwards
• Before you hand over a task, ask how costly it is to catch and fix
• Treat an agent like an employee, not an experiment

Tools and Frameworks Mentioned

• ⁠Claude Cowork⁠ - Anthropic's agentic harness, now available in the browser, that plans and executes multi-step work rather than just answering.
• ⁠n8n⁠ - Workflow automation builder with new native MCP support and human-in-the-loop approval steps, bridging automation and agentic AI.
• ⁠Model Context Protocol (MCP)⁠ — The open standard connecting agents to your key data and systems, now supported natively across major automation tools.
• ⁠Microsoft Azure CLI⁠ - A command line tool used with Claude Code to programmatically build agents and sandbox environments.
• ⁠OECD⁠ AI System Definition - The framework James uses to scope agents, defining an AI system by its degrees of autonomy and adaptiveness after deployment.

⁠Shape the show⁠

Connect with the Hosts:

• ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠James Gauci:⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ CEO of Cadent.

• ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Llew Jury:⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ Managing Director of Advancer at the AI Agency.

• ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Zero Shot:⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ Follow us on LinkedIn for clarity in a complex landscape.

  
  

🤠 Sponsor:

This episode is brought to you by ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Prompt Cowboy⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠, 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.⚡

Six months ago the AI leaderboard looked completely different. So who climbed, who slipped, and who fell off the board entirely?

In Episode 28 of ⁠⁠Zero Shot⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ , ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠James Gauci⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ and ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Llew Jury rebuild their AI tier list from scratch, ranking the major model providers across the US, China, and beyond. This is not about telling you which shiny tool to buy. It is about handing you a way to look at any AI company and instantly know whether it is right for your business. They break down the three ways a company can win, enterprise, distribution, and consumer, then rank every major player and close on how to actually choose without chasing the leaderboard.

Key Highlights

• The three ways to win: enterprise, distribution, and consumer
• Anthropic won enterprise and now leads on valuation
• Google is quietly winning distribution across phones and workspace
• ChatGPT still owns the consumer market, but the throne is wobbling
• Chinese open weight models now rival the best US frontier models
• Cost per task and token spend matter as much as raw capability
• Build redundancy, never depend on one provider staying switched on
• Match the winner to your needs, not the headline S tier
• 
  

The Full Tier List (July 2026)

S Tier: Anthropic (Claude)
A Tier: Google (Gemini), OpenAI (ChatGPT / Sol), Moonshot AI (Kimi K3)
B Tier: Nvidia, Alibaba (Qwen), Z.ai (GLM 5.2), DeepSeek
C Tier: Minimax
Split decisions: AWS lands at C for Llew, D for James. Microsoft lands at E for Llew, F for James.
Dropped off the board: Apple, one to watch for a self-hosted comeback once RAM prices fall.

Tools and Frameworks Mentioned

• Claude by Anthropic - S tier this round, leading enterprise adoption with its agent harness, Opus 5, and Fable 5.
• Gemini and Gemma 4 (Google) - A tier, strong on distribution, long context, and open weight smaller models.
• Kimi K3 (Moonshot AI) - A tier, a Fable-class reasoning model available as open weights on Hugging Face.
• Qwen 3.7 (Alibaba) - B tier, dominant in self-hosting though its top Max model is now closed weight.
• AWS Bedrock and Nova Models - Referenced for cheap, hardware-optimised inference within the AWS ecosystem.

Connect with the Hosts:

• ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠James Gauci:⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ CEO of Cadent.

• ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Llew Jury:⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ Managing Director of Advancer at the AI Agency.

• ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Zero Shot:⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ Follow us on LinkedIn for clarity in a complex landscape.

  
  

🤠 Sponsor:

This episode is brought to you by ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Prompt Cowboy⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠, 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.⚡

Most people with access to the most powerful AI model available are using it at about two percent of what it can do. What would you do with the rest?

In Episode 27 of ⁠Zero Shot⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ , ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠James Gauci⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ and ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Llew Jury ⁠ demystify Fable 5, Anthropic's most capable frontier model, and the class of model that now includes OpenAI's Sol and Moonshot's Kimi K3. This is not a hype reel. It is an honest, practical breakdown of what Fable actually unlocks, the finance and consulting work it can now do overnight, why it sometimes overthinks itself into hallucinating, and how to use it responsibly without burning through your token budget in an afternoon.

Key Highlights

• Why Fable is built for persistent, thorough, long-running looping tasks
• The mental shift: forget what you assumed AI could not do
• Real use cases: overnight financial reports that took analysts weeks
• Higher effort is not always better, max effort can trigger hallucination
• Cost per task is the metric that now matters most
• Top and tail big jobs with a powerful model, workhorse in the middle
• Fable hands off sensitive domains like cybersecurity to Opus 4.8
• Build redundancy, keep a fallback model ready if Fable goes dark

Tools and Frameworks Mentioned

• Claude Fable 5 - Anthropic's most capable frontier model, built for orchestration, persistence, and deep multi-step work, now available on Max and Teams premium plans.
• Kimi K3 (Moonshot AI) - A 2.8 trillion parameter open weight model challenging the frontier at roughly eighty percent of the cost, with open weights targeted for release.
• ChatGPT with Sol 5.6 - OpenAI's flagship high-power model in the Fable class, benchmarked closely against Fable and strong on image generation.
• Cost Per Task - An emerging evaluation metric comparing dollar cost across models, useful for managing total cost of ownership as token prices climb.
• RCTF Framework (Ep 26 Zero Shot) - Role, Context, Task, Format prompt structure, essential for getting quality output from a model this powerful.

Connect with the Hosts:

• ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠James Gauci:⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ CEO of Cadent.

• ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Llew Jury:⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ Managing Director of Advancer at the AI Agency.

• ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Zero Shot:⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ Follow us on LinkedIn for clarity in a complex landscape.

  
  

🤠 Sponsor:

This episode is brought to you by ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Prompt Cowboy⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠, 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.⚡

Why does AI give some people sharp, useful output and everyone else a generic, average answer?

In Episode 26 of ⁠ ⁠Zero Shot⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ , ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠James Gauci⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ and ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Llew Jury ⁠ go deep on prompt engineering, the skill that separates people getting real value from AI and people quietly disappointed by it. The honest truth is it is almost never the model. The models are astonishing. It is the brief. This episode covers why prompts underperform at a technical level, the four pillars of a strong prompt, and how everything changes when you move from a single prompt to designing the full context environment of an AI agent.

Key Highlights

• Why weak output is a briefing problem, not a model problem
• How a prompt reshapes the model's internal landscape of weights
• Making the implicit explicit is now a core professional skill
• The four pillars: role, context, task, format
• Set the role at world-class level to lift the output
• Less context done well beats more context done poorly
• The four parts of an agent brain: short memory, long memory, instructions, knowledge
• Prompting an agent is continuous improvement, not a one-time write

Tools and Frameworks Mentioned

• RCTF Framework (Advancer) - Llew's prompt structure covering Role, Context, Task, and Format, designed to be printed and reused across a team.
• ChatGPT Work with Sol - OpenAI's flagship high-power model in its new work mode, reviewed live against Fable 5 for leadership briefing tasks.

• Hermes - An agent harness that prompts and improves itself between sessions, tuning its own skills and memory as a continuous improvement loop.
• Retrieval Augmented Generation and Data Lakehouses, The shift from classic RAG toward trusted single-source-of-truth architectures like Snowflake and Databricks for real business context.

Shape the show: https://zeroshot.com.au/shape

Connect with the Hosts:

• ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠James Gauci:⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ CEO of Cadent.

• ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Llew Jury:⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ Managing Director of Advancer at the AI Agency.

• ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Zero Shot:⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ Follow us on LinkedIn for clarity in a complex landscape.

  
  

🤠 Sponsor:

This episode is brought to you by ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Prompt Cowboy⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠, 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.⚡

How do you tell the difference between a task worth handing to AI and one that should stay human?

In Episode 25 of Zero Shot⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ , ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠James Gauci⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ and ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Llew Jury tackle the decision most businesses get wrong: choosing what to actually point AI at. The instinct is to reach for the shiny, visible task: customer service, the website, email, but that is rarely where the value lives. This episode gives you a practical way to sort any task in your business into three piles: hand it to AI, keep it human, or a bit of both. Five fast triggers to spot the real candidates, seven questions to run each one through, and an honest look at the cost of automating the wrong thing.

Key Highlights

• Why your gut instinct to automate is usually the wrong target
• Automating a broken process just makes bad permanent, faster
• Failed early automations burn organizational trust and kill momentum
• The five triggers: repeat, sentence, dread, messy input, apology
• Let the AI do the dread work, not just the grunt work
• Seven questions that separate worth it from should we
• In regulated sectors, accountability is the governance question that matters
• The should conversation is where the real value lives

Tools and Frameworks Mentioned

• Claude Fable - Anthropic's high-capability model, restored last week in a stripped-back format outside the US, referenced for producing report and website work that would take an analyst weeks.

• Gemma 4 (Google DeepMind) - Google's open source model, cited as the kind of on-hardware model deployed in high-classification defense settings where public models are not permitted.
• The REWIRE Framework (Advancer) - Advancer's method for AI transition, moving through Reveal, Envision, and Weave to test viability and feasibility before building.
• System 1 and System 2 Thinking - Daniel Kahneman's model, referenced by James to frame when to let AI steer a fast call versus when deep human judgment is required.

Connect with the Hosts:

• ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠James Gauci:⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ CEO of Cadent.

• ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Llew Jury:⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ Managing Director of Advancer at the AI Agency.

• ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Zero Shot:⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ Follow us on LinkedIn for clarity in a complex landscape.

  
  

🤠 Sponsor:

This episode is brought to you by ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Prompt Cowboy⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠, 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.⚡