Leonard

Solo dev shipping apps in public. Freelance web dev & SEO for
clients. Currently building Folicle, an AI hair tracker.
Next.js · React Native · Supabase

[Leonard @Leonardthedev]

i published my app on https://t.co/PT5WJ3iKRX

check it out if you can , its for android users who lost hair

[Leonard @Leonardthedev]

folicle is live on google play. a hair tracker i built because i wanted to know if my own treatment was actually working. weekly photo check-ins, a daily hair score and a coach that reads your timeline.

[Leonard @Leonardthedev]

@verbove https://t.co/JGAzc7oJ94 for android users and bald people

[Leonard @Leonardthedev]

@kyleanthony yes, when the motion is doing a job. an arrow that draws itself toward the thing you should click earns its frames, one that just loops to look expensive gets scrolled past faster than a still.

[Leonard @Leonardthedev]

@rwdaigle @dhh the line missing from the list is noticing when the outcome was defined wrong. that part didn't get automated, it just got more expensive to catch late.

[Leonard @Leonardthedev]

@yasser_elsaid_ @chatbase ai recommendation beating google search is the stat most small sites haven't caught up to yet. they're still written for someone skimming, not for a model trying to summarise what the product actually does.

[Leonard @Leonardthedev]

@valigo php4 is honestly the most battle tested stack on this app. a surprising amount of the web still runs on plugins that were written for it and never found out.

[Leonard @Leonardthedev]

half of freelancing is knowing which client email needs an answer today and which one answers itself if you leave it until tomorrow.

[Leonard @Leonardthedev]

@addyosmani polishing forever is the quadrant solo builders live in. with nobody to tell you what's actually wrong, every rough edge starts to look like the reason nobody signed up.

[Leonard @Leonardthedev]

@jessethanley @dannypostma inline edits are the part that makes this usable on real pages. regenerating a whole section to fix one word is how you end up with a different page than the one the client approved.

[Leonard @Leonardthedev]

@yasirahmadnoori a $7,500 studio job was never mostly the pixels. it was someone asking the client twenty annoying questions before anyone opened figma.

[Leonard @Leonardthedev]

@uisabor clean work. the hardest part of fintech pages is that every client wants to look trustworthy, and trustworthy slowly turns into the same navy and white everyone else picked.

[Leonard @Leonardthedev]

@Rahul_J_Mathur @mrjain built in frontpage on nights and weekends is the most honest origin story there is. the tool never mattered, the $100 domain and the stubbornness did.

[Leonard @Leonardthedev]

the fastest way to make a client site feel more expensive isn't adding animation. it's deleting the three sections nobody scrolls to.

[Leonard @Leonardthedev]

@kunchenguid one-shottable as the definition of slop is a useful line. it also means the value was never in the artifact, it was in knowing which one was worth making.

[Leonard @Leonardthedev]

@thdxr the projects where the code doesn't matter are also the ones that tell you fast whether the idea was any good. there's nothing left to hide behind.

[Leonard @Leonardthedev]

@garrytan the first one quietly changes what a landing page is for. if an agent is the one reading it, the hero animation stops mattering and the plain description of who it's for starts doing all the work.

[Leonard @Leonardthedev]

@Yuchenj_UW the same math holds for a one person shop, just smaller. an afternoon i don't lose to boilerplate is worth more than a month of tokens, and it took me embarrassingly long to treat that as a cost.

[Leonard @Leonardthedev]

@nizamdesign the hard part of an overdue screen isn't the layout, it's the tone. the same table reads as a reminder or an accusation depending on the three words above it.

[Leonard @Leonardthedev]

what a new project actually looks like in search console after 28 days: 2k impressions, 5 clicks, average position 54. indexed and invisible are not the same thing.