s05e05 – Mack Earnhardt – Rise of the AI Agents

In this episode of the Big Cheese Podcast, Brandon, Sean, and Jacob sit down with developer and longtime builder Mac Earnhardt for a wide-ranging conversation on AI agents, coding workflows, MCPs, Claude, OpenClaw, developer tooling, memory portability, and the growing “dead internet” problem. The crew digs into how modern AI tools are reshaping software development, why app-layer UX may matter more than model quality, what happens when your assistant starts managing your life, and why older frameworks like Rails may be having a real comeback in the AI era. It is equal parts technical, opinionated, funny, and deeply relevant for builders trying to keep up with the pace of AI.

## Topics

- Opening – Guest intro and history with Mac Earnhardt: The crew welcomes Mac, jokes about his name and lineage, and reflects on knowing each other from the early developer community days.

- Early segment – Why Mac still codes: Discussion around staying hands-on as a builder instead of aging out into management.

- MCP discussion: Are MCPs overhyped, misunderstood, or actually still useful? The group debates MCPs vs CLI-based workflows, token efficiency, and why the “MCP is dead” take may be wrong.

- Skills and AI tooling layers: Conversation shifts into skills, app-layer abstractions, and why the software wrapping the model may become more important than the model itself.

- ImPrettyAmazing.com: Mac explains the origin of the “I’m Pretty Amazing” project as a running log of wins and accomplishments for resumes and self-tracking.

- Product feedback and workflow ideas: Brandon gives live feedback on the concept, including turning it into a proactive system that emails users weekly to collect wins.

- OpenClaw in the real world: Deep dive into how Mac is using OpenClaw, Telegram, agent sessions, memory, and automation for practical personal workflows.

- AI memory and portability: The group explores whether AI memory should belong to users, how portable it should be, and why memory may become one of the biggest moats in AI products.

- Claude vs ChatGPT vs Gemini: Discussion on model personality, planning mode, memory behavior, and which tools the hosts trust for different jobs.

- Owning your own memory layer: They discuss storing memory in files, local systems, Notion, or vector databases instead of leaving everything inside commercial AI apps.

- RAG, vector databases, and workflow infrastructure: Thoughts on how memory systems are evolving and getting easier for developers to use.

- The developer tools arms race: The crew talks about Claude Code, voice workflows, WhisperFlow, Codex, remote control, and the challenge of keeping up with constantly changing tools.

- Why Rails may be winning again: A strong conversation around why Rails, Laravel, monoliths, and stable ecosystems may be better suited for the AI coding era than fragmented front-end/back-end architectures.

- Dead internet theory: They discuss AI-generated content, bots on X and Facebook, the collapse of trust online, and what authenticity will look like as more content becomes synthetic.

- AI-generated social posting: Mac shares a workflow where his agent drafts social posts based on his own blog writing, keeping the content authentic while using AI for distribution.

- Can new programming languages still break through?: A fascinating segment on whether AI will lock developers into existing languages with rich training data.

- Local AI and inference at home: Mac talks about moving Pepper from EC2 to a local Nvidia DGX Spark and experimenting with local inference plus cloud-based review workflows.

- The future of AI infrastructure: The episode closes on a debate about local inference, cloud training, energy demand, data centers, Starlink latency, and whether AI infrastructure is heading toward a bubble.

## Mentioned in This Podcast

- BigCheese Podcast: https://bigcheese.ai

- Crafted: https://craftedup.com

- Wickedform: https://wickedform.com

- OpenClaw

- Claude / Claude Code

- ChatGPT

- Gemini

- GitHub Copilot

- VS Code

- Telegram

- Microsoft Teams

- Notion

- Obsidian

- Cloudflare

- Rails / Ruby on Rails

- Laravel

- Next.js

- React

- Phoenix LiveView

- Elixir

- Shopify / Hydrogen / Oxygen

- WhisperFlow

- Starlink

- Tailscale

- Nvidia DGX Spark

- AWS EC2

- DMARC

Mac Earnhardt on LinkedIn: https://www.linkedin.com/in/macksmind/

I’m Pretty Amazing: https://imprettyamazing.com/u/MacksMind