I Have 65 Tokens Left and Apparently That’s a Personal Problem

Let me explain what happened...

I found a vibe-coding tool for Mac called Glaze.

I had 120 free tokens....I had a mildly annoying workflow problem involving prompts and clipboard management and too many browser tabs and about eight thousand open Notes documents I never look at. I told the AI to build me a small floating window. That’s where things got away from me.

Here’s How the Evening Actually Went

The original ask was pretty simple. Make me a floating sidecar window that stays above other apps. Multiple tabs. Text blocks I can copy with one click.

I was thinking twenty minutes. Maybe thirty. The AI built it. It worked. And then I had a thought...

“The window is kind of annoying to move around.”

Done.

“Brand this. Red. My logo. PromptPocket.”

Done.

“When a node has nothing in it, show a Paste button.”

Done.

“When a node is just a URL, show a Launch button.”

Done.

“Put a mic button on each node so I can dictate.”

Done.

“When someone clicks the app name, open an About card with my info and a link to my site.”

Done.

I had not planned any of these things. Each one took about three minutes. Each one led directly to the next one. This is exactly how I end up in situations. At some point I looked at the token counter, looked at the clock, and realized I had done the thing I always do. I went in to look at something. I walked out having shipped software.

What PromptPocket Actually Is Now

It’s a small frosted-glass window that pins above everything on your Mac. Every Space. Every full-screen app. There when you need it, tucked wherever you put it.

Inside: tabs. Inside the tabs: nodes.

A node is just a block of text.

Could be a prompt. Could be a reusable email opening. Could be a snippet of client context. Could be code. Could be an instruction you’ve typed into twelve different AI sessions because you keep forgetting to save it. Could be a URL.

That last one now gets a Launch button. Empty nodes get a Paste button. Every node has a mic. Tap it. Talk. Done. The whole thing persists between sessions, remembers which tab you were on, and can be resized from a skinny sliver to something more substantial depending on the workflow.

Tap the title and you get an About card with my signature and consulting tagline. Because if I’m going to accidentally ship a product, I’m at least going to make sure it points back to me.

The Honest Part About Tokens

The honest part is that a “20 minute build” became considerably more than 20 minutes once I started adding things. That’s not a complaint. It’s an accurate description of how inference pricing works when you are constitutionally incapable of leaving things alone. Every iteration costs tokens. I knew this going in. I kept going anyway.

What this actually taught me is that the economics of software creation just went through a phase change. It used to cost developer time. Developer time is expensive and scarce. So the idea had to justify the cost. The cost filtered what got built. Now the cost is tokens. Tokens are cheap and refillable. The filter is different.

The question is no longer whether the idea justifies six months of development. The question is whether the idea justifies the next few minutes and the next few credits. PromptPocket would never have passed the first filter. It sailed through the second one.

What Glaze Is Actually Doing

After I got curious about the .app file size, I started poking at what Glaze was generating under the hood. The application isn’t a traditional compiled Swift binary. It’s something closer to a lightweight manifest that runs on top of a shared Glaze runtime. The common infrastructure is already installed. Which means Glaze is doing something architecturally significant. It is, essentially, the browser model applied to desktop software.

Before browsers, every web application would have needed to carry its entire networking stack, rendering engine, and JavaScript runtime inside itself. The browser solved that by putting all the common machinery in one place and making web applications lightweight artifacts that run on top of it. Glaze is trying to do the same thing for personal Mac software. Install the runtime once. Build lightweight apps against it. Share them through a store. Sound like anything you’ve used before?

The App Store Missed Something

Apple’s App Store changed the distribution problem. Before it, getting software meant downloading something from a sketchy website, clicking through six dialogs, and wondering whether your computer was about to join a botnet in Moldova. The App Store made discovery and installation clean and trusted. That was a genuine improvement. But it didn’t change who was allowed to make things. Developers made apps. Everyone else filed feature requests that went nowhere.

AI vibe coding on a shared runtime platform is a different kind of change. Not distribution. Production. When the cost of creating a working application drops below the cost of explaining why you need it approved, things that were previously below the economic threshold of traditional software start to exist. The App Store democratized distribution. This might actually democratize creation. Those are different things. One of them is bigger.

What I Keep Thinking About

The Glaze founder has called this an iTunes moment for software. I think that’s close but not quite right. iTunes was still a distribution play. The artists still had to make the music.

What I think is actually happening is closer to what YouTube did to television. Before YouTube, producing video meant equipment and editors and budgets and distribution deals. The cost filtered everything. So television aimed at mass audiences because it had to. YouTube collapsed the production cost so far that a channel for 600 left-handed woodworkers in the Pacific Northwest not only made sense — it thrived.

The software equivalent of that channel is PromptPocket. Too small to build commercially. Worth building personally. Useful to maybe a few hundred people who work with LLMs all day and have the same clipboard problem I had. Software that would never exist in any other model.

Anyway

So that’s how I ended up with 65 tokens left and a published Mac app that has a frosted glass interface, speech dictation, URL detection, an About card with my actual signature on it, and a store listing. I went in to see what the tool did. I came out with a product. The product is not going to make me a billionaire. It is also genuinely useful. Both of those things can be true, and in the emerging model of personal software, both of those things being true simultaneously is exactly the point.

The old question was: is this big enough to build? The new question might just be: would this make things a little less annoying? PromptPocket passed that test easily. Whether I still have the discipline to stop tinkering with my 65 remaining tokens is a different question entirely.

I will report back 😂

Don't have a mac?

I got you...and had to burn tokens elswhere 🤷‍♂️ so I dropped the Glaze chat in to Base44 for an instant clone ... its even a PWA (Progressive Web App) and stores all data locally. It should work on anything with a modern browser.

---

Ad: Use code: RICH99 for a discount.

Rich Washburn is a technologist and strategist working at the intersection of AI, infrastructure, and capital. He is Managing Partner and Chief AI Officer at Eliakim Capital and CIO of Data Power Supply.