If I wasn’t tweaking FontCrafter, I just spent the better part of a week with the OnePlus 15. Before that, the Galaxy S26 Ultra. Before that, the usual carousel of flagship phones that promise everything and deliver 90%.
And I keep landing on the same conclusion: the last 10% is where every company falls apart. Not on the hard stuff — the chipsets are fast, the screens are gorgeous, the cameras are (mostly) great. They fall apart on the stuff that should be easy. The stuff that only breaks if nobody on the team actually used the product before shipping it.
A clipboard notification you can’t dismiss. Rounded screen corners so aggressive they clip your own app’s footer. A folder system that arbitrarily limits you to 12 items per page when there’s clearly room for more. AI features that throw internal service errors. A task switcher that judders into place like it’s surprised you opened it.
These aren’t hard engineering problems. These are “did anyone live with this for a week?” problems.
This pattern isn’t limited to phones. It’s everywhere in tech right now. The product you’re using is almost certainly 90% great and 10% infuriating — and that 10% is almost always a product management failure, not an engineering one.
Someone decided to ship before the details were right. Someone prioritized a spec sheet bullet point over the daily experience. Someone built a feature that demos well in a boardroom but annoys you every morning.
Here’s the takeaway: stop chasing the perfect device. Start choosing based on which headaches you’re willing to live with. That’s not cynicism — it’s actually freeing. Once you accept that every option has compromises, you can pick the compromise that bothers you least instead of chasing something that doesn’t exist.
For me, software fluidity and customization win over camera quality. For you, it might be the opposite. Neither answer is wrong. The wrong answer is buying something expecting perfection and being frustrated when the 10% hits.
The companies that win the next few years won’t have the best specs. They’ll be the ones that close the last 10% gap. The ones whose product teams actually use what they ship. I’m still waiting.
That same frustration — the “did anyone actually use this?” frustration — is what pushed me to start building my own tools. Not by learning to program, but by describing what I wanted to an AI and having it build the thing for me.
I needed to Base64 encode something. Every online tool is covered in ads or wants my email. So I built Studio B64 — clean, fast, no nonsense.
I needed to shrink a video file. The options were “upload to our server” (no thanks) or figure out FFmpeg. So I built Video Shrinker — drag, drop, done. Everything stays in your browser.
I had a broken SVG file. If you’ve ever tried to fix mangled SVG code, you know. So I built SVGlyph — paste in the mess, get a working file back.
I wanted custom poker chips for game night. The design tools were either too limited or too complex. So I built MakerChip Maker — a 3D parametric designer that exports files ready for a 3D printer.
Every NFC app on Android looked like it was designed in 1998. NFC tags are great — tap your phone and instantly transfer info, like a QR code without the robot-barf image. Cheap, useful, fun. But the apps? Hideous. So I built NFC Maestro — and then vibe coded my first Android app in under an hour. I’m not a developer. I just refused to accept ugly as the only option.
I wanted to turn images into 3D-printable ASCII art. This one was just for fun. ASCII Forge exists because I thought it would be cool. It is.
The result: over 170 free browser-based tools at arcade.pirillo.com. No installs, no accounts, no data leaving your machine.
Beyond the problem-solving stuff, I’ve been having a ridiculous amount of fun building things I just wanted to exist:
FontCrafter — Turn your handwriting into a real, working font. This one blew up — 31K+ views on X, top 50 on Hacker News.
Kind Rewind — Generates photorealistic VHS tapes with custom cover art. Pure nostalgia.
Paper Jam — A dot matrix printer simulator. It sounds like the real thing.
Pin Toy — That pin impression toy from the ‘90s, but it uses your webcam.
Sakurascape — A cherry blossom scene generator. Just beautiful.
Be Leaf — A generative garden you can grow in your browser.
CoffeeTime — Photo mosaic art generator.
Morning Cockpit — A personalized morning dashboard.
Space Needle — A game. Because sometimes you just need to play.
Pocket Calculator — A beautifully designed calculator.
Spice Rack — A visual spice reference for your kitchen.
Duly Noted — A legal pad note-taking app that uses my actual handwriting as the font.
Glyph Extractor and Font Anatomy — Deep-dive font inspection tools.
Syncope Structure, Design Workshop, Website Workshop — Creative tools for different workflows.
Every one is free, runs in your browser, and requires no account.
I spend a lot of time exploring what people are building. Here’s what caught my eye lately:
Handy — Open-source, offline speech-to-text for your desktop. Press a shortcut, talk, and your words appear wherever your cursor is. Nothing gets sent to the cloud. It’s the kind of tool that should’ve existed years ago.
Jan — Run AI models right on your computer instead of relying on cloud services. Open-source, works on all platforms, keeps your conversations completely private. A ChatGPT alternative that lives on your machine.
SkillsMP — A searchable collection of hundreds of thousands of reusable “skills” for AI coding tools. If you’re curious about how the AI development ecosystem is maturing, this gives you a sense of the scale.
Paperclip — This one’s wild. An open-source platform for creating companies staffed entirely by AI agents — with org charts, budgets, task delegation, and audit trails. Already has 22,000+ stars on GitHub. Whether or not you want to run an AI company, it’s a fascinating look at where things are heading.
SuperPlan — A planning layer that sits between your ideas and your AI tools. Interviews you about what you want to build, writes a proper spec, then feeds structured tasks to your AI. The idea: AI doesn’t build the wrong thing if it starts with the right plan.
Claude Code Ideas — A project idea generator for people who want to build with AI but aren’t sure what to make. You tell it about yourself and it suggests personalized projects. Nice starting point if you’re curious.
TrustMRR — A database of real startup revenue numbers, verified through Stripe. No more inflated screenshots on Twitter. An eye-opener if you’re interested in what indie software businesses actually earn.
Pane — A desktop app for managing multiple AI coding sessions in parallel. If you’re the kind of person who ends up with 15 terminal windows open, this is for you.
Every workshop I host, one registered participant wins a $100 Amazon gift card. The last two winners:
🎉 The creator of Agent Sandy Safety — from our most recent virtual event where the keyword was “Agent.” Check out their creation here.
🎉 Emerald Stack — from our last hybrid event where the keyword was “Space Needle.” See what they built.
You don’t need to be technical to win. You just need to show up, participate, and build something — anything — during the workshop.
I want to make it ridiculously easy for anyone to publish what they create online. No hosting accounts, no deployment pipelines, no terminal commands. Just paste in your HTML and it’s live. I’m hoping to build this platform soon, because the biggest barrier to people sharing what they make isn’t the building part anymore — it’s the publishing part.
I host workshops called ctrl+alt+create live — hybrid events (in-person in Seattle + virtual) where we build real software together using AI. No coding experience required. The whole point is showing that anyone can do this.
If you’ve ever thought “I wish there was an app that did ____” — this is your chance to build it. 👉 Register here!
Yours Digitally,