Two AI assistants running on one Mac Mini, 24/7 (here's what happened)

Hi everyone,

Did you miss me last week? I missed you. I genuinely wonder if anyone notices when the newsletter doesn't show up. If you did, let me know, because it'll make my day.

A lot happened since we last talked:

Let me walk you through all of it.

I did it. I canceled ChatGPT Business and moved my team members to Claude Teams.

It wasn't some big strategic decision; we just…weren't working in ChatGPT anymore.

Claude became the go-to for 99% of tasks.

I’ve kept my ChatGPT Pro plan though, because GPT 5.3 Codex is still superior for certain things.

Starting next week, we’ll be running internal workshops to walk the team through Claude Chat, Claude Code, Claude Code Plugins, Skills, and Apps.

Why? Because the gap between what Claude can do and what most people know about it is massive. I want my team to close that gap.

My honest take: Claude is just unstoppable right now.

Here's something that would have sounded insane a year ago.

I have two AI assistants running around the clock on a Mac Mini in my office. Both connected to Telegram. Both doing actual work.

I have two Telegram folders on my phone: one for Mia, one for CLAWD. I message them like colleagues—two AI assistants working 24/7, one on Claude Opus 4.6, one on OpenAI Codex.

Two weeks ago, I delivered a 90-minute workshop for EBC (the European Bridge Chapter of EO). The focus: Claude Code.

These are business owners, not developers, so we started with Claude Code: learning the terminal, the commands, and the power of it.

Then I told them:

“If you're not as technical and you don't want to see the terminal, you can work in Claude Cowork today. Build your plugins, your skills, all from a visual interface. It's Claude Code, with a UI.”

Several attendees started testing Cowork during the session, and there was palpable excitement about what they could automate.

Good news: Claude Cowork is no longer Mac-only; it’s now available on Windows.

95% of AI pilots fail to scale.

Answer 8 quick questions to receive your personalized, actionable AI adoption roadmap in minutes.

Link below 👇

AI Readiness Assessment

A few tools have earned their spot in my daily workflow.

Granola is a top AI note-taker that I’ve mentioned before. Now it has MCP integration.

Here's what that means in practice: I finish a sales call, and Claude Code pulls the transcript directly from Granola. Then it creates the deal entry, updates tracking, drafts follow-ups. No more copying and pasting transcripts. It just happens.

I host prospect pages, proposals, and slide decks on Vercel subdomains with password protection. I can spin up a custom page for a prospect in minutes with Claude Code. It's fast and it looks professional.

I'm switching from Google Meet and Notion scheduling to Calendly and Zoom. It's the enterprise standard for webinars and cohorts, and my clients expect it.

Here's the ironic part of my week.

I'm rebuilding aioperator.com from Framer to Astro + Sanity CMS. Pure code. A real codebase. Version control. The works.

Why? Speed. Scalability. Content automation. I can now hook up Claude Code to my CMS and generate pages, blog posts, and landing pages directly.

But here's the funny thing.

Ten years ago, when I was head of marketing at a tech company, I took websites AWAY from developers. I pushed for WordPress with Divi and Elementor. Then Webflow. Then Framer.

"We don't need developers for this," I said.

And now I'm going back to pure code, because AI makes it possible for someone like me to build and maintain a real codebase.

I owe an apology to every developer I pushed aside ten years ago. You were right. Code is better. I just needed AI to catch up so I could work with it myself.

I won't name names, but this story is too good not to share.

One of my clients has a CTO who was skeptical of Claude Code. On our first call, he told me: "I'm happy for you to convince me."

Three weeks later, he sent me a message: "I'm 3-4x more efficient. Thank you."

If a CTO with decades of engineering experience is converted, this is real, not just hype for non-technical people. Claude Code is making experienced developers better at their jobs too.

Six months ago, I barely knew what a framework was.

Now I understand servers, deployments, Vercel, GitHub CI/CD, CLIs, MCPs, APIs.

Not because I took a course. Because I've been working alongside Claude Code every single day and reading everything it produces.

Here’s my biggest piece of advice for Claude Code (and vibe-coding in general):

Don't just accept the outputs.

Read it—all of it. Ask it to explain things. Switch to the exploratory mode and dig into the codebase. Treat it like a mentor, not a vending machine.

You don't need a four-year computer science degree anymore, but you do need curiosity and a willingness to read.

My background? Sound engineering, and a 10-week coding course that I found tedious.

I never became a developer, but now I create 20-30 page websites in an hour. I manage a full codebase. I deploy to production.

We're all becoming developers now. Whether we planned to or not.

One more thing: webinars are coming back!

The format will be bigger than before, less frequent, and with limited seats.

I don't have a link yet, but expect an announcement by the end of February.

Sign up right away, because these will fill up.

I wrote the Human + AI Manifesto months ago. Some days it felt aspirational. This week it just felt like my life.

We don't work with AI because it's trendy. We work with AI because it changes what's possible. Every tool, every workflow, every late-night experiment brings us closer to something that used to take entire teams.

It doesn't feel lonely, either. Working with Claude Code feels like working with a friend. A tireless, patient friend who never judges your questions.

Hit reply and tell me: what's the one thing you'd automate if you had an AI assistant running 24/7? I'm collecting ideas for the upcoming webinar series.

Best,

Tim