The SessionClick landing page at sessionclick.com was fine but felt a little dark and technical, and it had no demo video on it even though I shot one a couple of days ago. Time for a refresh.

Attempt one: too much, too fast

I asked Claude to look at the page and pitch a redesign. Its first proposal was a full rework — cream/paper background, a serif display font (Fraunces), a “bento” grid of feature tiles, the YouTube Short embedded in a phone-shaped frame in the hero. Claude described the direction, I said go, and it implemented the lot.

I looked at the result for a minute and reverted it via Git. The colour flip was too far from the app’s actual look, and the typography change made it feel like a different product. Actually it looked like an Anthropic website. Good thing everything is versioned.

That’s a useful pattern, by the way: let the AI try a maximalist version, see how it feels, then dial back. Cheaper than agonising over the brief up front.

Attempt two: small palette nudges + the video

I told Claude to keep the existing palette and fonts and just brighten things a touch, and to put the demo video on the page. That landed:

  • --bg lifted from #0f0f0f to #1e1e1e, surface from #1c1c1c to #2f2f2f, borders and text bumped up to match. Same theme, just less near-black.
  • The green stayed but went one notch brighter (#4CAF50 → #66BB6A) so it still pops against the lighter background.
  • A simple phone-shaped CSS frame holding the YouTube Short as an iframe (autoplay, muted, looped, no controls).

Then a couple of small adjustments:

  • I asked to bring the existing banner image back between two content sections — done.
  • Then asked to swap positions: banner back in the hero, video standalone in its own showcase section. Also done.

A new feature card: hands-free foot control

I’d added Bluetooth foot-pedal support to the app a few days ago but the landing page didn’t mention it. I asked Claude to drop the “Tap Tempo” card and replace it with a foot-control card, and to figure out the wording from my project notes.

This part was nicer than I expected. Claude grepped through my Obsidian vault, found the Perplexity research note on page-turn pedals, then went into the actual Android source and pulled up MainActivity.kt’s dispatchKeyEvent — saw exactly which key codes are bound (arrow keys, Page Up/Down, Space, Enter) and that they all trigger play/stop. From that, it wrote copy that’s specifically true: works with iRig BlueTurn, PageFlip, AirTurn, Donner, no setup. I didn’t have to dictate the marketing copy or check the implementation myself — it cross-referenced both.

Small bonus: a Hugo warning

The Hugo preview was logging found no layout file for "html" for kind "taxonomy". The site has no posts or tags, so the fix was to disable taxonomies entirely in hugo.toml:

disableKinds = ["taxonomy", "term"]

One line, warning gone, two empty /categories/ and /tags/ folders no longer generated.

Where the AI helped vs. didn’t

  • Me: the design direction (reject the maximalist version, keep the existing palette/fonts, just brighten), the layout decisions (banner in hero, video standalone, drop Tap Tempo, add foot control), and reverting via Git when I didn’t like something.
  • Claude: the over-ambitious first pitch, the smaller second pass, the CSS palette tweaks, the phone-frame embed, and — the part I liked most — investigating my notes and the Android source to write accurate copy for the foot-control card without me having to spec it.
  • Gemini: not involved today; this was the Hugo landing page, not the KMP app.

Today’s changelog

  • New: demo video embedded on the landing page in a phone-shaped frame.
  • New: “Hands-free foot control” feature card on the landing page.
  • Changed: brighter dark palette across the site; slightly more vivid green accent.
  • Fixed: empty taxonomy pages and the related Hugo warning.

Time spent today: ~45 minutes, most of it deciding what not to change.

This blog documents my attempt to build and ship a music app as a solo developer, with AI assistance. The AI does a lot of the work. I try to be specific about what.