Venture Step

You have to make decisions before the future is certain. The question is how much you can afford to get wrong. In this solo episode of Venture Step Podcast, I review Peter L. Bernstein’s Against the Gods: The Remarkable Story of Risk, exploring humanity’s evolving understanding of risk and what it means for entrepreneurs, business owners, and anyone making decisions with incomplete information. One hypothetical bakery connects the story. Before the first customer arrives, you’ve committed money to the lease, ovens, and ingredients. From there, we work through the questions that follow: which evidence should you trust, what could a loss actually cost you, and how do you keep one mistake from ending the business? The discussion moves from Roman games of chance to Pascal and Fermat’s interrupted wager, where dividing a $100 pot requires thinking about the games still to be played. John Graunt’s mortality records lead into the value of collecting data, the danger of biased samples, and why a predictable average across many businesses doesn’t tell one owner what their actual loss will be. Even a bakery’s product testing can mislead: friends praising your jalapeño croissants are different evidence from customers coming back to pay for them. Then we turn to utility and the St. Petersburg paradox. An attractive potential payout doesn’t settle whether a risk makes sense for the person taking it. A two-ship example explores how diversification can reduce the chance of losing everything while leaving smaller losses possible. The discussion of the 2008 financial crisis brings us back to a critical assumption: exposures that look separate may still fail together. Those questions become practical again when AI starts making decisions for a business. An instruction to maximize croissant production needs limits around equipment, spending, and employee wellbeing. Delegating the work doesn’t make its consequences disappear. I also explore framing effects and why “95% success” can land differently from “5% failure.” My own approach is to leave room to experiment and make mistakes while avoiding catastrophic losses. The episode closes by asking what we might give up if the pursuit of predictability leaves no room for people to defy the odds. What’s a risk you’re willing to take, and what downside would make it unacceptable? Share your perspective and follow Venture Step for entrepreneurship, industry trends, and the occasional book review. Further listening E15: Bet on Yourself, a review of Annie Duke’s Thinking in Bets, on decision quality, luck, and uncertain outcomes. E120: The Rise of the AI Coworker, on managing AI agents with clear goals, boundaries, and human review.

AI is getting more capable. What happens to the people using it? I revisit earlier Venture Step conversations about AI tutors, digital companions, and intimate deepfakes, connecting those concerns to reported harms today. I discuss loneliness, excessive chatbot validation, personal data, child safety, and the TAKE IT DOWN Act. My experiences studying for the CPCU, migrating podcast files, and tackling dbrand's Robot City puzzle show why I find AI useful. They also raise a question: who bears the consequences when a useful tool becomes a harmful influence? EARLIER EPISODES E35: AI tutors https://youtu.be/6BwWKkZ7aeA E61: Gemini Canvas https://youtu.be/qoGIyz0azww E67: TAKE IT DOWN Act https://youtu.be/JMf253z5VEY E28: Meta AI Studio and AI companions https://youtu.be/_zlpebV6cHY E33: Productivity and AI relationships https://youtu.be/nAW62_6pXaU E84: Ani, Meta Vibes, and digital addiction https://venturestep.net/episodes/the-digital-addiction-crisis-x-ai-s-ani-meta-vibes-and-the-pursuit-of-ai-slop Further listening, E106: Robert LoCascio on children and AI https://venturestep.net/episodes/kid-company-uare-ai-why-you-are-the-best-ai-with-robert-locascio REFERENCES John Oliver: AI Chatbots (2026) https://www.youtube.com/watch?v=Ykvf3MunGf8 Francesca Mani and Elliston Berry's Senate testimony https://www.commerce.senate.gov/meetings/field-hearing-take-it-down-ending-big-techs-complicity-in-revenge-porn/ FTC: TAKE IT DOWN enforcement and platform removal requirements https://www.ftc.gov/news-events/news/press-releases/2026/05/ftc-begins-enforcing-take-it-down-act OpenAI: Data use and training controls vary by product and settings https://openai.com/policies/how-your-data-is-used-to-improve-model-performance/ dbrand Robot City puzzle https://dbrand.com/shop/limited-edition/robot-city CPCU designation https://web.theinstitutes.org/designations/cpcu SUPPORT In the US, call or text 988 for crisis support. https://988lifeline.org/

Short summary Most AI demos stop when something appears on screen. This one starts there. Dalton Anderson uses OpenAI Codex to turn fictional customer feedback into a working dashboard, then critiques the first output, steers the redesign, and explains the system around reliable agent work: Plan as the map, Goal as the contract, skills as reusable procedures, and evaluations plus stop conditions for agentic loops. A practical episode for founders, operators, and builders who want better results from Codex without giving up judgment. ### Mobile-first show notes Most AI demos stop when something appears on screen. This one starts there. Dalton gives Codex a fictional customer-feedback dataset and asks for a decision-ready dashboard. The first output works, but it is not good enough. That becomes the real lesson. This episode shows how to move from a one-shot prompt to a workflow you can inspect, steer, and verify. You will learn: - Why Plan is the map and Goal is the contract - The four parts of a strong goal: outcome, context, constraints, and done-when evidence - When a repeatable workflow should become a skill - How clear names keep a growing skill library usable - Why every agentic loop needs an evaluation and a hard stop - Where human judgment still matters The demonstration uses fictional data. No outreach is sent. ## Official OpenAI resources - [Codex use cases](https://developers.openai.com/codex/use-cases) - [Build skills](https://learn.chatgpt.com/docs/build-skills) - [Build plugins](https://learn.chatgpt.com/docs/build-plugins) - [Subagents](https://learn.chatgpt.com/docs/agent-configuration/subagents) - [AGENTS.md](https://learn.chatgpt.com/docs/agent-configuration/agents-md) - [Follow a goal](https://learn.chatgpt.com/use-cases/follow-goals) - [Scheduled tasks](https://learn.chatgpt.com/docs/automations) - [Git worktrees](https://learn.chatgpt.com/docs/environments/git-worktrees) - [OpenAI Codex repository](https://github.com/openai/codex) - [OpenAI Plugins repository](https://github.com/openai/plugins) Note: the older [openai/skills repository](https://github.com/openai/skills) is deprecated and now directs readers to OpenAI Plugins. ## Skill repositories worth exploring - [Anthropic Skills](https://github.com/anthropics/skills): official Claude skill examples and templates - [Superpowers](https://github.com/obra/superpowers): a cross-agent software development workflow and skill collection - [Microsoft Skills](https://github.com/microsoft/skills): skills and custom agents for Microsoft developer workflows - [Microsoft Learn Agent Skills](https://github.com/MicrosoftDocs/Agent-Skills): Microsoft and Azure skills grounded in Learn documentation - [Gemini CLI](https://github.com/google-gemini/gemini-cli): Google's open-source coding agent with Agent Skills support - [Gemini CLI Agent Skills guide](https://geminicli.com/docs/cli/using-agent-skills/) - [Agent Skills specification](https://github.com/agentskills/agentskills): the open format behind portable skills - [Vercel Skills](https://github.com/vercel-labs/skills): a cross-agent CLI for discovering, installing, and sharing skills Install selectively. Read a skill before trusting it, understand the tools and permissions it can use, and test it on bounded work first. ## Chapters 00:00 Why this episode exists 01:45 Turning fictional feedback into a dashboard 02:29 Commands, context, compact, goals, and Plan 04:16 What the first plan is doing 07:05 Steering the build with butter yellow 09:03 Reviewing the first dashboard 10:37 Plan is the map, Goal drives the work 15:54 An honest review of the redesign 17:23 Building a Goal and the four-part prompt formula 21:08 Commands, skills, and reusable workflows 24:57 Naming skills so they stay usable 26:34 Broad threads and focused projects 28:27 A simple context-and-constraints analogy 30:52 Keeping agentic loops safe 32:21 Closing

Exploring the Power and Limitations of AI Imagery and Voice Models In this episode, Dalton Anderson dives into a hands-on exploration of various AI models, including image, voice, and interactive storytelling tools. He shares real-world demos, highlighting their strengths and areas for improvement, especially from a marketing and content creation perspective. Whether you're curious about how AI can enhance your creative projects or want a behind-the-scenes look at the current state of AI models, this episode offers valuable insights.KEY TAKEAWAYS: Microsoft’s AI models in beta offer the ability to generate images, transcribe speech, and produce voice outputs, providing a versatile toolkit for creators. The Google Nano Banana Pro image model excels at creating realistic, lived-in scenes with detailed aging and environment, but can sometimes produce inconsistent details. Prompt specificity is crucial; AI-generated images can misinterpret vague instructions, leading to hallucinations or illogical outputs, such as unrealistic headgear without appropriate suits. Voice models demonstrate strong emotional range, with certain voices like Harper’s delivering particularly convincing storytelling, while others struggle to evoke specific emotions like anger or sadness. Interactive story features currently limit depth due to character caps and simplistic responses, making them less effective for complex storytelling or nuanced character interactions. Microsoft AI Platform: microsoft.ai Venture Step website: https://www.daltonanderson.net/venture-step/ Call to Action: Subscribe to Venture Step for more behind-the-scenes founder stories and AI insights. CHAPTER TIMESTAMPS:00:00 - Introduction and overview of AI models showcased00:28 - Microsoft’s AI beta platform and available models00:56 - Demonstration of image generation and critique from a marketing perspective01:25 - Comparison between Google Nano Banana Pro and Microsoft’s image model02:28 - Visual demo of Microsoft’s UI and image outputs03:01 - Creating a futuristic diner scene and assessing visual emphasis03:33 - Analyzing inconsistencies in AI-generated scenes04:33 - AI depiction of Gemini with futuristic elements and critiques05:51 - Visual analysis of the emphasis on key objects like shakes and burgers07:04 - Halucination in AI images—cat paw coming out of a table08:02 - Detail critique: gears and aged objects in AI art09:30 - Evaluation of realistic details like butterflies and aged books10:08 - Google’s attempt at capturing a dusty carpenter’s workshop11:04 - Comparing Google’s and Microsoft’s realism in craftsmen scenes12:23 - Limitations of AI in facial expressions and lighting realism13:00 - Exploring voice-to-text capabilities and story creation14:14 - Generating a short story with emotional beats15:47 - Testing voice model emotion: Harper, Ethan, Olivia16:33 - Switching perspectives to the Ant Queen in storytelling21:12 - Creating emotional stories from different viewpoints24:29 - The superiority of Harper’s voice for storytelling25:29 - Challenges with interactive stories and character limits28:34 - Wrapping up with a futuristic AI story titled 'Neon Gears of Hunger' CONNECT & LINKS: Microsoft AI Platform: microsoft.ai Venture Step website: https://www.daltonanderson.net/venture-step/ Call to Action: Subscribe to Venture Step for more behind-the-scenes founder stories and AI insights. Music licensed through Soundstripe. Code: OWITVHYLRLGCXG5V, SZ1PQVE2IOUP0XDN

This episode explores the evolving relationship between Microsoft and OpenAI, the strategic shifts in AI model development, and the competitive landscape shaping the future of artificial intelligence. Dalton Anderson provides an in-depth analysis of key events, investments, and technological advancements. keywords AI, OpenAI, Microsoft, AI models, tech competition, AI investment, cloud computing, AI strategy, GPT, Anthropic key topics Microsoft and OpenAI partnership history AI model development and competition Strategic investments and infrastructure in AI Impact of leadership changes and corporate politics sound bites "Microsoft is hiring top AI talent." "The future of AI is highly competitive." "The AI landscape is rapidly evolving." chapters 00:00 The Evolution of Microsoft and OpenAI's Relationship 10:10 The Crisis: Sam Altman's Ouster and Its Implications 19:54 The Shift: From Partnership to Competition 30:06 The Future Landscape of AI: Competition and Innovation resources OpenAI Official Website - https://openai.com Microsoft AI Models - https://microsoft.com/ai Anthropic - https://anthropic.com Google DeepMind - https://deepmind.com Sam Altman Twitter - https://twitter.com/samaltman Elon Musk XAI - https://x.ai Google AI - https://ai.google AWS AI - https://aws.amazon.com/ai Oracle Cloud - https://oracle.com/cloud SoftBank Vision Fund - https://softbank.com learn more https://www.daltonanderson.net/venture-step/