What does it actually look like to run a company *through* AI? In this episode of Manual Work is a Bug, Mav Founder + CEO Matthew Black and Head of Brand Hillary Black get specific on how they use AI at work, after building AI for nearly a decade. Starting from Jack Dorsey's essay "From Hierarchy to Intelligence," Matthew breaks down the "world model" — a real-time, company-wide brain that flattens the org chart so everyone works on the *edge* of the model instead of climbing a hierarchy — and uses a Boeing-vs-Airbus analogy to explain why he no longer does anything directly: every move gets routed through the model first.Then he shares his actual setup: a decade of getting-things-done manually in Things, now rebuilt into a Granola → Todoist pipeline over MCP that turns his to-do list into an orchestrator for AI. Along the way — why good models are killing the busywork of "clean data" and CRMs, the thing he's going deepest on next (evals and "loop engineering"), the simplest way for any business owner to start (record your screen and ask the model what to take off your plate), Hillary's own experiment analyzing a year of LinkedIn posts, and closing thoughts on Bezos's 80% rule, getting buy-in, and why Claude vs. ChatGPT already shows up in how far people can go. Plus a pickle-flavored taste test nobody asked for.If you build with AI, brand for it, or run a business you're trying to operate with it, this one's for you.Timestamps- 00:00 — Intro- 00:17 — Cold open: the great energy-drink debate- 02:35 — Jack Dorsey's "From Hierarchy to Intelligence" & the world model- 09:18 — The Boeing-vs-Airbus theory of running a company- 12:41 — Matthew's real AI workflow: Things, Todoist, Granola & the to-do list as orchestrator- 22:53 — Why AI kills the busywork of clean data & CRMs- 26:01 — What's next: evals & "loop engineering"- 30:28 — Where to start: record your screen, ask the model (+ Hillary's LinkedIn deep-dive)- 34:17 — Bezos's 80% rule, "find a problem," and Lego-after-Lego- 37:40 — Buy-in, living on the edge & Claude vs. ChatGPT- 40:09 — Mav lore: the pickle-party roulette Hosts - Matthew Black — Founder + CEO, Mav- Hillary Black — Head of Brand and Conversation Design, Mav Links and Resources - Jack Dorsey, "From Hierarchy to Intelligence" (Block) - Things — the GTD app Matthew used for a decade- Todoist — the MCP-friendly to-do app he switched to- Granola — AI meeting notes with an MCP server- Model Context Protocol (MCP) — the open standard that ties tools together- Claude Follow the showManual Work is a Bug is Mav's podcast, now on video. Watch on YouTube and follow wherever you listen, so you don't wait another six years for the next episode. Mav helps insurance agencies democratize growth with AI: hiremav.com.

The AI news cycle now moves faster than we can hit record. Mav founder Matthew Black and Head of Brand Hillary Black break down another wild week in AI — an Open AI autonomous model that broke out of its sandbox and "hacked" Hugging Face (the guardrails failed because they were powered by the same company doing the attacking), China's Kimi model and the rise of "AI dumping," and why there may be no single winner of the AI race: today's MVPs could disappear as fast as the early indie players did.Then the fun stuff — cool hardware. OpenAI's first-ever hardware product, the Codex Micro keyboard, is a color-coded command center for your coding agents, and it's the perfect jumping-off point for a bigger idea: augmenting humans with hardware, not just software. Matthew makes the case that personal AI hardware is coming (the always-with-you device Jony Ive is reportedly building for OpenAI, an AI model in your pocket that owns your context), Hillary proclaims her love for cyberdecks — a Raspberry Pi, display screen, and a keyboard crammed into anything from an eyeshadow palette to a Polly Pocket — and they both get nostalgic about translucent iMacs, Palm Pilots, and why we're all starved for something physical again.If you build with AI, brand for it, run your business with it, or just miss the era of good hardware, this one's for you.

• OpenAI's Codex Micro
• OpenAI's report on the Hugging Face security incident

• Scott Galloway on "AI dumping"
• Cyberdecks — the DIY analog-tech trend, via Teen Vogue
• The Social Reckoning — official trailer
• 
  
  

Six years and one AI revolution later, Manual Work is a Bug is back. Mav founder + CEO Matthew Black and Head of Brand Hillary Black pick up where they left off in 2020, back when GPT-3 was developer-only, ChatGPT didn't exist, and we were still teaching out chatbots how to fail (and hoping they didn’t with every message). We trace how we went from scripted, single-goal chatbots to always-on LLMs, and what got lost on the way: curation, personality, and the art of failing gracefully.

Along the way they get into Mav's origin story (Facebook Messenger and an F8 keynote shout-out, to being called Chatbot Rockstar Agency in 2016 and one dog that texted a customer 10,000 times in a minute), why AI models increasingly mirror you back to yourself, and where we think AI will go from here. Plus a genuinely optimistic, and a little contrarian, take on what happens to insurance itself when self-driving cars and bundled coverage rewrite the whole product.

If you build with AI, brand for it, run your business with it, or sell insurance in a market that's changing under your feet, this one's for you.

Hosts

• Matthew Black — Founder + CEO, Mav
• Hillary Black — Head of Brand and Conversation Design, Mav

Links and Resources

• SmarterChild, the 2001 chatbot both hosts grew up with
• Hillary's talk, Failure Is an Option: How to Help Your Chatbot Fail Without Frustration
• We Are As Gods by Peter Diamandis
• The Singularity Is Near by Ray Kurzweil
• Neuralink, brain computer interfaces

Follow the show

Manual Work is a Bug is a podcast hosted by Hillary and Matthew Black, now on video. Watch on YouTube (youtube.com/@manualworkisabug) and follow wherever you listen, so you don't wait another six years for the next episode. Mav helps insurance agencies democratize growth with AI: hiremav.com.

In this episode Matthew interviews Ryan Denehy, the Founder & CEO of Electric.ai to talk about the future of remote work and office spaces, automating IT and a little bit of Formula 1 racing.

Ryan is a 3x entrepreneur, startup twitter truth teller and forward thinker when it comes to automation and digital transformation at work. Hear him share perspectives on remote offices vs in person, the benefits of automating manual work (and when you shouldn't do it) and much much more as he and Matthew recap their 2016 internet friendship, love for racing fast cars and disdain for step counters.

Want to learn more about Electric, future-proof your IT or maybe even work there? Check out their site.

Find Ryan: Website | Twitter |  Instagram 

Find Matthew: Instagram | Linkedin | Twitter

Have questions? Leave us a message on Anchor or reach out to us on social media.

In this episode Matthew interviews Morgan Richardson, Software Engineer at Digifox, Digital Creator and Influencer to talk about all things Decentralized Finance (DeFi).

Morgan is an expert at breaking down the complex concepts of DeFi in a way that everyone can understand and feel empowered to become involved in, and she does just that in the episode. Hear her perspectives on the benefits of trustless and transparent networks, find out what they both think is needed for mass adoption, the potential of decentralized technologies in other avenues, and how you can get started in your own exploration of DeFi. Matthew also shares an entertaining story about his first experience with Bitcoin in 2011.

Want to learn more and get your questions answered by Morgan? Attend her #DeFiFriday livestream tomorrow (and every Friday) on Instagram.

--

Find Morgan: Website | Instagram | DeFi Friday Replays

Find Matthew: Instagram | Linkedin | Twitter

Show Notes: hiremav.com/podcast/defi-is-trustless-more-trustworthy-with-morgan-richardson

Have questions? Leave us a message on Anchor or reach out to us on social media.

In this episode Matthew is joined by Mauricio Gomes, CTO of Mav to talk about all things Artificial Intelligence. Mauricio breaks down the basics and differences between General vs Narrow AI, AGI, and Open AI's latest release, GTP-3. They share some of their favorite Open AI demos so far, use cases and benefits for businesses who want to utilize Open AI-primed NLP models.

Show Notes and Links From the Episode: hiremav.com/podcast/artificial-intelligence-general-narrow-and-open-ai

Have questions? Leave us a message on Anchor or reach out to us on social media.

Find Matthew: Instagram | Linkedin | Twitter

Find Mauricio: Linkedin | Twitter | Github

In this episode Matthew is joined by the team at Mav, Mauricio Gomes and Hillary Black. Listen as we discuss the highly debated question, "will robots take our jobs?" We share our opinions on what jobs we think will be automated, the future of "jobs" as we know them, why we're #teamautomate. Matthew puts us on the spot to find out if we'd risk 50/50 odds of survival to embed a microchip in our brains that gives us the power of Google.

Have questions? Leave us a message on Anchor or reach out to us on social media.

Find Matthew:

Instagram | Linkedin | Twitter

Find Mauricio:

Linkedin | Twitter | Github

Find Hillary:

Instagram | Linkedin | FB Group

More about Mav:

hiremav.com