I sat down this week to count how many things I’ve built with AI over the last couple of years. I assumed it would be around fifteen. Maybe twenty if I was being generous and included things that exist only on my laptop and possibly in my imagination.
It is over fifty.
Fifty apps, plugins, sites, and small experiments. None of which I would have built without AI, because before this, most ideas like these would have stayed exactly where they started: in a notebook, a voice memo, or a slightly overexcited conversation in the pub. AI changed that. It turned “someone should build this” into “hang on, I think I can.”
That part has changed.
The Pootle apps
The first cluster is the six Pootle apps.
PootleWriter, a distraction-free Markdown writer that publishes straight to WordPress.
PootlePhotos, a drag-and-drop image uploader that does exactly one thing.
PootlePlayground, a visual builder for WordPress Playground blueprints.
PootleSites, which runs multiple WordPress sites entirely in your browser.
PootleBooks, for publishing small web-based books.
And PootleNoodle, a thing that captures guitar riffs straight into a WordPress post, because I have a band and a problem.
I built all six in less than a day for each one. A few years ago this would have been a full developer team, several months, and a frankly upsetting invoice.
The bigger builds
Then there’s Tomorrow Times. An idea I’d carried around for fifteen years — a prediction market dressed up as a satirical newspaper. I built the whole thing in a weekend. AI-generated illustrations, real-time notifications, custom emails, leaderboards, an admin dashboard. Not a prototype. A working app. I still don’t really believe it was possible to build such complexity using AI alone.
DR-01 is an ambient synth that runs in your browser. Two sound layers, an arpeggiator, drone mode, looper, effects. I built it in a single Claude conversation, mostly because I wanted to stop pretending I could play the Korg Monologue properly.
jamiemarsland.com is my personal site, built as a headless WordPress setup pulling the content via the REST API. (Connecting Claude to WordPress over that same API takes five minutes, by the way.)
scrappy.world is a digital scrapbooks app — I wrote up the moment it turned into something more.
There’s also a notes app I built, a newsletter signup page built with Bolt and Omnisend, and a WordPress Playground generator that creates personalised websites from a name and a message.
WordPress plugins
And then there are the plugins.
Write, which strips WordPress down to a blank page and a cursor.
Write or Die, which deletes your words if you stop typing for five seconds, because apparently we need that now.
The Front-End Editor, so non-technical teams can edit content without ever seeing wp-admin.
Sticky Header for Block Themes.
An Unlisted Posts plugin.
Scrubber, which I built with Cursor.
What this actually means
Let me be clear about what these things actually are.
Tomorrow Times is a real product. Database, authentication, a payment-grade email system, AI image generation, RSS ingestion, real-time notifications, leaderboards, web search for resolutions, an admin dashboard, a full mobile experience. It would have been a six-figure build a few years ago and a year of someone’s life.
DR-01 is a playable instrument. Not a toy. You can sit with it for half an hour and make something genuinely beautiful, with effects, looping, drone layers, and presets that completely reshape the sound.
PootleSites runs multiple full WordPress installs in your browser with persistence. PootlePlayground builds Blueprints visually. The Front-End Editor solves a real problem that has annoyed agencies for fifteen years. Write or Die is small but it works exactly as it should. scrappy.world is a finished thing people can use today.
These aren’t demos. They are working software.
What’s changed isn’t whether the work is impressive. It’s the friction between having the idea and the thing existing.
The idea-to-thing distance used to be enormous. You’d think of something at breakfast, write a spec by lunch, brief a developer in the afternoon, get a quote on Friday, and by the time the thing existed you’d lost interest and started a podcast instead.
That whole chain has collapsed. You think of something. You describe it. It exists. You refine it. It gets better. Sometimes the first pass is rough, and that turns out to be fine, because you can just keep going until it’s not.
The interesting question isn’t whether AI lets you build things. It clearly does, and the things are real.
The interesting question is what you choose to build when the cost of building has effectively gone to zero.
I keep accidentally answering it.
I can’t stop building apps with AI.
Leave a Reply