1,000 People Showed Up. Here’s the Story, What’s Changing, and a Giveaway.

From ChatGPT-4 running a business in 2023 to 1,000 subscribers today

This is a different kind of post. Not a tutorial, not an experiment. Just me, being honest about where this newsletter came from and where it’s going.

---

Yesterday, Digital Thoughts hit 1,000 subscribers.

I sat with that for a moment. Then I opened my laptop and started typing, because that’s what I do.

1,000 is not viral. It’s not a hockey-stick story. But for me it means something real, so I want to mark it properly. Not with a confetti graphic or a “thank you for your support” paragraph that sounds like a press release. Just the actual story.

Where this started

About three years ago I published my first Substack post. It was about whether ChatGPT-4 could run a small business. Not a thought experiment. I actually tried it. Set up a simple product, gave the AI a budget and a brief, and watched what happened.

It worked. Kind of. We made some money, got some traction. It was messy and interesting in equal parts.

The technical setup was embarrassing by today’s standards. The context window was 4,000 tokens. (If that number doesn’t mean anything to you: imagine trying to run a business with a collaborator who forgets everything after a ten-minute meeting.) I was building “universal memory keys” to pass context between chat sessions manually. I had automated ChatGPT-4 with Zapier so it could actually do things, not just suggest things. It was clunky. It also kind of worked.

That era shaped everything I write about. The core question was always: can AI create real value, or is it just a very impressive autocomplete? I still think that’s the right question. I’m still testing it.

What Digital Thoughts actually is

I’ve described this newsletter a hundred different ways. The most honest one: it’s a lab notebook.

I build things with AI, I break things with AI, and I write about it. Sometimes the results are good. Sometimes the AI builds exactly what I ask for and it’s still completely wrong. That’s the honest version of this stuff.

I’ve tried local models on my MacBook and iPhone, because I wanted to see how far you can get without cloud infrastructure. The gap is closing faster than most people think. I’ve compared AI coding tools until I landed back where I started. That journey was its own lesson. I’ve written about why 88% of companies say they use AI but only 6% see real results.

I also care about things beyond tools. Education. The future of work. What happens to office jobs when AI stops being a productivity add-on and starts being the default worker. These are not abstract questions for me. I have a kid starting school soon. I think about this stuff a lot.

The newsletter will keep covering all of it. No narrowing to “just AI tools” or “just automation.” That would be more legible, maybe, but it would also be a lie about how I actually think.

What’s being built right now

Three things I’m actively working on that will keep showing up here:

My AI agent is getting more capable. I’ve been building Wiz (my personal AI agent) for over a year. It runs on my Mac, handles scheduling, writes, deploys, and improves itself overnight. I’m also building a local version for Mac and iPhone that works without internet. The architecture is getting interesting and I’ll document more of it here.

The open source question. A lot of you have asked whether I’ll open-source Wiz. Honest answer: I want to. The problem is it’s so customized to me that a raw release would not be useful to most people. I’d need to build a more universal version first. That takes a pause in the endless self-improvement cycle, which is hard to justify when things are working. But I haven’t given up on it. When it happens, you’ll know.

Project Money is getting more interesting. The experiment started the same way as all of this: can an AI agent extract real value from human ideas and turn it into products people actually pay for? It’s working. Not perfectly, not at scale, but the direction is clear. The next frontier I want to explore is physical products. Not me designing them, but me having an idea and the agent handling the rest: production, graphics, store setup, customer feedback loop. I think this is more possible than it sounds.

The 1,000 subscriber offer

The store at wiz.jock.pl has 16 products right now. Total value if you bought everything: over $800. That’s playbooks, templates, automation systems, agent blueprints. Things I actually use and built as byproducts of the experiments here.

A yearly subscription to Digital Thoughts costs $129 and gets you everything in the store, plus everything I add going forward. If you subscribe this week, I’m doing 30% off for 7 days.

This includes everything in the store ($800+ value) and everything added going forward. Price locked in permanently.

I’m not here to pressure anyone. Everything I write is free. The paid tier exists for people who want more finished versions of the experiments: the system, not just the story.

The giveaway

I use Claude Max, the highest plan at $200/month. It gives you access to the most powerful models at a level that’s hard to describe unless you’ve used it. I go to places with it I wouldn’t think were possible six months ago.

For the 1,000 subscriber milestone, I’m giving one month of Claude Max to one person.

How to enter:
0. Subscribe to Digital Thoughts

1. Leave a comment on this post answering one question:

“What’s one thing in your work or life you’d automate if you could?”

That’s it. One comment. I’ll pick a winner based on most creative answer in 7 days and announce it in the next post(and in comments). No purchase required.

I’m genuinely curious what people answer. It’ll probably turn into future content, because that’s how this place works.

Thank you

1,000 people chose to let me into their inbox. That’s not nothing. I don’t take it lightly.

The deal is simple: I keep experimenting, I keep writing honestly about it, and you get to follow along and do something with the ideas if they’re useful. That’s what this has always been and what it’s going to keep being.

See you in the next one.

Pawel