Next session is Saturday, June 27 at noon Pacific, and the fireside chat is all about making your own desktop apps. Grab a free seat here!
Here’s a fraction of what you missed at our May event:
On Monday, June 8, Apple stood on stage at WWDC and told a billion iPhone owners they can now make their own Safari extensions by describing what they want. No code. No developer account. No Xcode. You type “a button that saves this recipe for later,” and Safari builds the extension and drops it into your toolbar. The feature is named, with zero subtlety, “Describe an Extension.”
In the same keynote, Apple gave the Shortcuts app the same treatment. You describe an automation in plain language, by typing or by speaking it out loud, and Apple Intelligence assembles the steps. The example Apple demoed: “Notify my partner when I leave work with my ETA.” Behind that one sentence sits a location trigger, a Maps lookup, and a Messages action, wired together without the person ever seeing a single block.
You describe what you want, the AI writes the software, you guide it and ship it. We’ve watched non-technical people build working apps in a single afternoon with that exact loop. And now the largest hardware company on the planet has baked the loop into the operating system that runs in your pocket. It’s happening!
Someone at Apple decided “say what you want” is the interface now, and pointed it at everything they could.
For a year, the pushback on vibe coding has come in two flavors. The first: “That’s a toy, real software is harder than that.” The second: “Sure, a hobbyist can throw something together, but no serious company would ship that way.” Both arguments lean on the same hidden assumption, which is that describing software in plain language is a novelty that lives at the edges, not a real interface that belongs at the center.
Apple just moved it to the center.
When Apple puts “describe what you want and we’ll build it” into Safari and Shortcuts on every supported iPhone, iPad, and Mac, the question stops being “is this a real way to make software.” The question becomes “what do you want to make.”
Notice what Apple is betting on here. They are betting that the average person, the one who has never opened a code editor and never will, has ideas for small pieces of software and has been blocked from making them by syntax. Remove the syntax, and the ideas come out. That is the entire thesis of this community.
If you want to understand why this matters beyond Safari, look at Shortcuts, because Shortcuts has been carrying this exact problem for a decade.
Shortcuts is powerful. You can automate tasks across all your apps, process information, manage files, control your smart home, chain a dozen steps into one tap on your home screen.
And almost nobody uses it.
For years it sat there as a power-user toy, because building a shortcut meant opening a blank editor and understanding actions, variables, inputs, and outputs before you could make the simplest thing work. The capability was never the barrier. The blank editor was. People with real automations in their heads took one look at the interface and backed out, and a decade of updates never fixed that, because the updates kept adding power to a tool whose actual problem was the front door.
Natural-language creation moves the front door. Instead of staring at an empty canvas trying to reverse-engineer which blocks you need, you say “remind me to grab an umbrella when tomorrow’s forecast shows rain,” and the app assembles a draft you can adjust inside the same editor. The hands-on coverage from the beta is honest that the draft won’t be perfect and the more ambitious automations still need cleanup.
Starting from a working draft instead of a blank page is the whole difference between a tool the technical few mastered and a tool a regular person will open.
That’s the same wall every non-technical person hits with every creative tool ever made. Take away the interface tax and the demand that was always there comes pouring out. Shortcuts is about to run that experiment on a scale no automation tool has ever had, and I would bet real money that usage climbs in a way ten years of incremental polish never produced.
Apple’s “Describe an Extension” makes small, bounded, personal tweaks to web pages. A save button. A widget that strips clutter off a page. A control that changes how one site behaves for you. It lives inside Safari’s sandbox, it runs on your device, and it’s scoped to the web. Shortcuts is similar in spirit: you describe an automation, and it chains together actions that already exist across your apps. These are personal tools, built for one person, solving one annoyance.
What we do at this workshop has a higher ceiling. We’ll help you build whole apps - like a maze generator that turns a photo of your kid into a printable puzzle. Each one is a standalone thing with its own logic, its own design, and its own life on the open web (or desktop) that anyone can use.
Apple’s version is a personal screwdriver. Ours is a full workshop with various tools.
When you vibe code an app from scratch, there’s no catalog and no fence. If the capability you need doesn’t exist yet, you describe it and it gets built. That’s a different kind of making, and it’s why a person with no parts catalog at all can still ship a calculator that nobody had written before.
But the move underneath both is identical. In both cases, the human describes the outcome in plain language, and the machine handles the implementation. The skill being exercised is “do you know what you want, clearly enough to say it.” Writing the code stopped being the hard part. Apple just declared that skill mainstream and handed the tooling to everyone who owns their hardware.
When the same core interaction shows up in a weekend hobbyist’s browser app and in Apple’s flagship operating system, that’s an interface winning.
The first thing the AI gives you is rarely the last thing you ship. The work is in the conversation after the first draft, the part where you look at what came back, notice what’s wrong, and say “no, more like this.” Describing well is a skill, and like any skill it rewards practice.
Apple putting a “describe it” box in front of a billion people doesn’t make everyone good at describing. It makes everyone able to start.
There’s a geography catch worth knowing too. The new Apple Intelligence features powering this, including the Safari and Siri pieces, are not shipping in the European Union or China when the betas roll out. And none of it is final yet. The developer beta is out now, a public beta lands in July, and the real release comes in September alongside the new iPhones. So if you go looking for “Describe an Extension” on your phone today and can’t find it, that’s why. The thing is coming, not here.
I point these out not to deflate the news but because pretending a new tool is magic is how people get burned. The honest version is more useful: this is real, it’s powerful, it’s not finished, and the person who learns to drive it well will get far more out of it than the person who expects it to read their mind.
The fear is that AI makes people passive, that the machine does the thinking while the human nods along. In practice I see the reverse. The person describing a Safari extension has to get clear on what they want before the AI can build a thing, and getting clear is more active and more demanding than clicking through a menu someone else designed.
The tool refuses to move until you’ve done the part only you can do, which happens to be the part worth getting good at.
If describing what you want is the skill that suddenly matters, it’s worth saying out loud that it’s learnable. Here’s what I’ve watched work with people who walked into the workshop sure they couldn’t do it.
Describe the outcome, not the steps.
The instinct from years of using computers is to spell out every click. The machine doesn’t need your clicks. It needs to know where you’re trying to end up. “Notify my partner when I leave work with my ETA” is an outcome. Figuring out the location trigger and the Maps lookup is the machine’s job, and it’s good at it.
The perfect first prompt is a myth that mostly wastes your time.
Keep a running list of the small things you wish your software did. Most people have trained themselves not to notice these, because for their whole lives the answer was “too bad, you’re not the developer.” Start noticing again. Every “I wish this page had a button that...” just became a thing you can have by the end of a sentence.
If you’ve been on the fence about whether learning to build with AI is worth your time, Apple just answered the question for you with a roadmap - and, even if you don’t use Apple products, I’m confident you can expect these types of features to land natively on other platforms.
I was suggesting platforms would have this type of functionality baked in almost a year ago - and, happily, I wasn’t wrong about that.
When the interface to software becomes language, the bottleneck shifts. For decades the gate was “can you code.” That gate kept out teachers, nurses, small-business owners, artists, retirees, kids, anyone with a problem and no patience for semicolons.
The new gate is “do you know what you want.” That’s a far more human gate, and far more people are already standing on the right side of it. They just haven’t been told they’re allowed through.
Step back far enough and this is the third time the same thing has happened. The graphical interface let people use a computer without memorizing commands, and the number of computer users exploded. The web let people publish without owning a printing press, and the number of publishers exploded. Each time, a layer of required expertise got absorbed into the tool, and a population that had been locked out walked in and changed what got made.
Plain-language software creation is the next layer coming off.
The syntax and the structure that used to live in your head move into the machine, and what’s left for you is the part that was always the real point: deciding what should exist.
Most people haven’t felt the size of that yet. We’re early enough that “I made an app” still sounds like a brag instead of a Tuesday. Give it a year of Apple’s version sitting in a billion pockets, of people casually building a Safari button or a weather Shortcut over coffee, and the strangeness wears off. Making a small piece of software stops being something you hire out and becomes something you do, the way taking a sharp photo stopped being a profession the moment the camera in your pocket got good enough.
I’ve spent the past year as a one-person proof of that. I can’t write code, but I’ve shipped 330+ free apps anyway (every one of them built by describing what I wanted to a browser tab and refining things until they worked). I’m not special. I’m just early. What Apple announced this week is the on-ramp for everyone who comes next, and there are a lot of them.
The tools caught up to the idea. Now it’s a question of who picks them up.
What: The next live vibe coding workshop! A fireside chat, then about two hours of building with AI, then we show what we made.
When: Saturday, June 27, noon Pacific. Runs roughly two hours.
Fireside chat: how to make your own desktop apps.
Where: Foundations, 1605 Boylston Ave, Seattle, and live online.
Cost: Free (and we take donations)
Bring: A charged laptop and an idea, or just an idea. We’ll give you a creative starting point if you arrive empty-handed.
Pre-register now: luma.com/4o27yap0
New here? Start with what vibe coding is, then come build something.
Six new ones since the last issue, all free, all running entirely in your browser, none of them collecting a thing about you.
Maze Maker. A printable maze generator that builds grid, hex, triangle, or circular mazes, or shapes one from any photo or emoji. Solve it on screen, then export to PDF, PNG, or SVG with an answer key. A free alternative to the watermark-covered maze sites.
Coffee Counter. Answer six quick questions and it matches you to one of 55 espresso, iced, and cold-brew drinks, shows you what’s in it, and tells you how to order it without guessing. Yes, this is the one that Claude Design basically one-shotted at our most recent event.
Markdown Galley. Paste or drop a Markdown file and export a clean, selectable, print-ready PDF, with a library of document styles and a live preview. A free alternative to fiddling with a paid Markdown-to-PDF tool.
Spool Story. If you’re into 3D printing, this one’s for you. Drop in an STL or 3MF and it tells you the filament grams, length, and cost, plus how much of your spool one print eats. It surfaces the number your slicer keeps to itself, and you can now flip on an electricity toggle to add real power cost on top. A free alternative to keeping a filament spreadsheet you’ll never update.
Force Finder. Never seen Star Wars and not sure where to jump in? Answer a few questions and it plots your entry point: where to begin, which order to watch, what to put on tonight.
Skill Cartridge. Embed an AI skill file inside the pixels of a PNG, share that single image, and load it back to pull the skill out. Built the way PICO-8 games are passed around, where the picture is the cartridge.
No tool roundup this issue. Instead, here’s what the community built and demoed at the last session, each one made live by someone who described an idea and watched it come to life. Different people, different brains, one Saturday:
A Startup Accelerator Game by Nicole Doyle
Exploring Cultures Through Food by Yoel Weldeselassie
Local vs. Cloud by Aleksander Allen
Uncovering Dark Patterns by Mehar Mira
Watch a couple! They’re each about a minute long (or so). It’s what you can expect to either make or see at the next event you register for…
A few things rattling around in my head from the past couple weeks:
Give a room of people the same Saturday and the same tools and they still scatter in every direction, except when a smell of “coffee” (as that event’s shared keyword) is in the air. There’s a product lesson in there about how much of “what should I build” is just “what’s in front of me right now.”
The Safari extension feature solves a problem most people didn’t know they were allowed to have. For years the answer to “I wish this website did X” was “too bad, you’re not the developer.” A whole category of small daily annoyances got filed under not-my-job. Watch how fast that category empties out once the fix is one sentence away.
The hard part of describing software is knowing what you want. The describing is the easy half. Most of building anything is discovering that the thing you asked for wasn’t the thing you needed, and the only way to find that out is to hold a wrong version in your hands.
A person who builds one small tool for themselves rarely stops at one. The skill compounds and the appetite grows. Apple is about to find out what happens when a billion people taste that for the first time.
Yours Digitally,
Comments
Nothing yet. Say the first thing.
Sign in to join the conversation.