The Fizz

Justin Mannhardt & Kellan Danielson

Two AI practitioners. One conversation. No script. Justin Mannhardt and Kellan Danielson are the co-founders of Tiny Bottle AI, a consultancy built on one idea: focus beats coverage. Every week on The Fizz, they break down what's actually happening in AI and share an unfiltered look at building their company from scratch. The Drip — The week's biggest AI stories, filtered through a practitioner lens. Not a news recap. The stories that actually matter, why they matter, and what they mean for people doing real work with AI. Inside The Bottle — A deeper look at what Justin and Kellan are building, shipping, and learning at Tiny Bottle AI. Client patterns, tools they're testing, decisions they're making, mistakes they're catching. The stuff that usually stays behind closed doors. Lab Notes — Quick hits from the field. What each of them worked on, tested, or discovered that week. Short, punchy, standalone value. New episodes drop Weekly. Subscribe to the companion newsletter, The Weekly Fizz, at tinybottleai.com/the-fizz.

  1. Oct 21 ·  Video

    AI Agents Are Working. Can We Keep Up?

    00:00 AI agents are working. Can we keep up? 00:44 Meet Greg Petrossian 01:37 From prompting to agents that keep working 03:31 One AI agent or a whole team? 08:07 Can AI spot what you missed? 10:04 More AI output, more dropped balls? 13:39 Could you work a week without AI? 15:28 What happens when your AI expert leaves? 20:06 AI doomerism vs. everyday reality 25:47 When AI agents break the rules, who is responsible? 29:18 Can AI defense keep up with AI attacks? 34:40 Why AI policy keeps falling behind 36:35 Keeping AI useful when the headlines get weird 39:14 Would you let an AI agent spend money? Episode 27: AI Agents Are Working. Can We Keep Up? Justin and Kellan compare notes with Greg Petrossian on how they use AI agents: one primary assistant, specialist agents, and an agent that delegates to others. They discuss what’s working, where they still need to make the decisions, and what happens when they take on more work than they can manage. In This Episode: Greg’s software factory: architect, engineer, security, QA, and red-team roles.Why Kellan still measures his AI assistant against a great human assistant.An agent catching the wrong event time, and why one success doesn’t prove it’s reliable.Learning the work before handing it to AI, taking on a manageable workload, and documenting what the next person needs to know.Greg’s experiments with agent payments and invoicing, and the risks of letting agents handle money.New episodes every Wednesday. Subscribe to The Weekly Fizz newsletter at https://tinybottleai.com/the-fizz/subscribe Music licensed through Soundstripe. Code: B3XRCWG555MV6MUT

  2. Oct 14 ·  Video

    Your AI, Your Responsibility

    00:00 Your AI Can Do That. Should It? 02:39 Encourage, Require, Refuse: A Simple AI Policy 03:13 Purpose: Chase Business Wins, Not AI Hype 05:16 Protection: What Your AI Must Never Expose 08:55 Permission: What Can AI Do Without Asking? 12:04 Shadow AI: Give People a Better Option 16:00 Private AI Models: Keep Your Data In-House? 18:00 Could “No AI” Be a Business Strategy? 19:31 Where Human Craft Still Matters 21:20 Should AI Write to Your Customers? 23:09 Judgment: Convincing AI Can Still Be Wrong 25:20 Ownership: Your AI, Your Responsibility 27:48 Learning: Why AI Governance Can’t Be Set-and-Forget 30:15 Turn AI Policies into Enforceable Rules Episode 26: Your AI, Your Responsibility What should people be able to do with AI at work? Justin and Kellan walk through six governance components and a simple way to answer each: we encourage, we require, we will not. Along the way, they discuss agent permissions, the work customers value, and who remains accountable for AI output. Purpose, protection, permission, judgment, ownership, and learning.Connecting AI use to business improvement, including what to do with the time it saves.Why a person’s access doesn’t automatically determine their agent’s authority.The tradeoffs of a deliberate no-AI strategy and protecting the craft customers value.Matching review standards to the stakes and involving someone who knows the subject.Revisiting policies as the tools change, then checking which controls can actually be enforced.New episodes every Wednesday. Subscribe to The Weekly Fizz newsletter at tinybottleai.com/the-fizz. Music licensed through Soundstripe. Code: MXB6GUDAJNZEHPCJ

  3. Oct 7 ·  Video

    Stop Starting from Scratch with AI

    00:00 AI Vendor Updates and Tool Parity 05:13 Challenges with AI Memory Management 08:10 Building Trust and Onboarding in AI Tools 09:36 Building Skills for Daily Planning 18:17 Enhancing Productivity with AI 21:49 Integrating Loops and Schedules 24:48 Establishing Decision Boundaries with AI 27:17 Navigating AI Trust and Boundaries 28:42 Understanding AI Models and Their Decision Frameworks 31:11 The Future of AI: Productivity and Skills Development Episode 25: Stop Starting From Scratch with AI A useful AI chat is one thing. Having a routine that knows how you work is another. Justin and Kellan walk through how context, skills, schedules, loops, and human judgment fit together, using their own daily planning routines as the example. The Drip: Microsoft Copilot's latest agent features and the growing overlap between Copilot, Claude, and ChatGPT.OpenAI's dots and the move toward always-on agents.Why Justin keeps durable context outside individual tools, and what happens when a memory from one task applies everywhere.Inside The Bottle: Start with work worth repeating: the routine stays recognizable even when each day's inputs change.Bring in the context the task needs, through connectors or the information you're comfortable supplying.Turn the way you work into a skill, then improve it through actual use.Use a schedule to have the preparation waiting when you sit down.Two kinds of loops: check for changes, or keep working toward a defined result.Decide where human judgment belongs. A routine status update and feedback on a client's proposal call for different boundaries.New episodes every Wednesday. Subscribe to The Weekly Fizz newsletter at tinybottleai.com/the-fizz. Music licensed through Soundstripe. Code: 4GEUDSHZMWPHJJZO

  4. Sep 30 ·  Video

    Your AI Opinion Needs Experience

    00:00 Live from St. Paul MN 01:11 The Drip: AI’s capability overhang 03:23 Why leaders need firsthand AI experience 06:58 Building vs. planning: two different approaches 14:10 How changing models change your workflow 16:37 Research blind spots and AI gaslighting 21:28 Plausible answers and strategic judgment 24:44 Trust, risk, and accountability 27:40 Finding useful business applications 31:06 Building your own informed opinion Episode 24: Your AI Opinion Needs Experience How much of your opinion about AI comes from using it? Recording together in St. Paul, Justin and Kellan compare the habits they’ve changed through hands-on work: where they give agents more room, where they stay methodical, and what still needs their own judgment. The Drip: Ethan Mollick’s “The Overhang” and the gap between AI capability and everyday use.Why a successful demo and a dependable business process are different things.Building an informed opinion through actual work, not just headlines.Inside The Bottle: A reversal in approach: more room for agents when building, more deliberate research when developing strategy.Asking for competing points of view before deciding what you think.Research blind spots—and the value of reading beyond an AI-generated answer.Why growing confidence in AI does not remove responsibility for checking its work.Mentioned in the conversation: The Overhang — Ethan MollickAI Pathfinder — answer 10 questions about your business to request a free, human-prepared Pathfinder Brief on where to focus and what to do first.New episodes every Wednesday. Subscribe to The Weekly Fizz newsletter at tinybottleai.com/the-fizz. Music licensed through Soundstripe. Code: T32XXQNFPVWDQ5YD, 5UT98UGHD683QC45, 9CVESBUKPVWTGG5K

About

Two AI practitioners. One conversation. No script. Justin Mannhardt and Kellan Danielson are the co-founders of Tiny Bottle AI, a consultancy built on one idea: focus beats coverage. Every week on The Fizz, they break down what's actually happening in AI and share an unfiltered look at building their company from scratch. The Drip — The week's biggest AI stories, filtered through a practitioner lens. Not a news recap. The stories that actually matter, why they matter, and what they mean for people doing real work with AI. Inside The Bottle — A deeper look at what Justin and Kellan are building, shipping, and learning at Tiny Bottle AI. Client patterns, tools they're testing, decisions they're making, mistakes they're catching. The stuff that usually stays behind closed doors. Lab Notes — Quick hits from the field. What each of them worked on, tested, or discovered that week. Short, punchy, standalone value. New episodes drop Weekly. Subscribe to the companion newsletter, The Weekly Fizz, at tinybottleai.com/the-fizz.