Can you build luxury software?

_{Day 178/100}

Hey—It's Tim. 

It’s finally raining in Vancouver again.
Feels like being back in the Norf (north of the UK above the “welcome to the north” sign). IYKYK

I’ve been thinking a lot about design - how the app feels and what makes something feel luxury.

There’s a “little luxuries” trend right now (the sequel to “little treats”). And, whilst it’s mostly consumer goods, it got me wondering:

Is there such a thing as luxury software?
What would that look like? What would that feel like?

If “little luxuries” are aesthetic, performance, and experience - what does that mean for software? Let’s break it down.

Aesthetic (looks so good it disappears)

Great software gets out of the way.

When I was drafting buildings, I lived inside SketchUp. Compared to 3ds Max or Blender it felt barebones - on purpose.

The blankness was a feature. I modelled faster because there was nothing to fight.

It got out of the way.

What this looks like:

• Ruthless white space, strong type, zero chrome. States (empty, loading, error) are designed on purpose.

• Motion with meaning: 150–240ms eases; overshoot under 4%. Never “cute.” Always clarifying.

• Iconography and color that communicate, not decorate.

Who nails it (and why):

• Linear - typography + spacing so precise you forget it’s a web app.

• Arc - opinionated surfaces that still fade behind the page.

• Claude - As much as I’m on the OpenAI train right now. Claude just looks better. Serif fonts get me every time.

Performance (so fast it feels inevitable)

Alibaba made it trivial to pump out passable products. Margins raced to zero. Quality followed. Consumers learned to spot drop-shipped junk and started paying more for things that actually work.

AI did the same to features and content. Slap “AI-powered” on the box, route every click through a giant model, ship some shimmer and a spinner, and call it innovation.

Output got cheap. Experience got worse.

Yes, I can make 5,000 blogs - but at what cost (cough—ByWord—cough)?

Luxury performance is the counter-move:

• LLM last, not first. Use native code, heuristics, and precomputation for 90% of interactions. Call a model only when it creates non-linear value.

• Perceived speed is real speed. Sub-100ms for non-AI core flows. For AI flows: optimistic UI, prefetch on hover, and layout-matched skeletons so 800ms feels instant.

• Determinism > vibes. Guardrails, evals, and fallbacks. If the model hesitates, degrade gracefully to a fast, good-enough path.

• Cache like you mean it. Prompt templates compiled, retrieval warmed, results memoized per user/context. No one should wait for the same answer twice.

• Zero theatre. Kill fake “typing…” animations and novelty bounces. Replace with progress you can trust

It doesn’t just “do the thing.”
It does the thing better than anything.

Who nails it (and why):

• Superhuman — keyboard-first; emails feel like keystroke combos, not chores.

• Figma — multiplayer at 60fps; “heavy” work that feels weightless.

Experience (feels like cheating - in a good way)

Luxury is the feeling you’re always on the right track. The app anticipates you, forgives you, and celebrates small wins.

What this looks like:

• Onboarding that gives you a “first success” in 60 seconds. Time-to-joy is a KPI.

• Micro-feedback that lands (tiny haptic nudge, subtle confetti only on meaningful moments).

• Opinionated defaults; fewer choices, better outcomes.

Who nails it (and why):

• Tella - Makes Loom feel cheap. Just does what Loom used to do, without all the bloat. And I enjoy using it.

So… can you charge a premium?

Yes - if you own all three: aesthetic, performance, experience.

“Looks nice” isn’t a business model. “
Feels inevitable and makes me faster” is.

CFOs still ask “why the price?”

A 20-minute luxury audit (do this today)

• Kill one control. Remove a field, a step, or a button from your most-used flow. Ship the simpler path.

• Measure time-to-joy. From sign-in to first meaningful success. Get it under 60s.

• Prefetch the future. Hover prefetch, optimistic updates, skeletons that mirror final UI.

• Tune motion. Standardize durations/easings; ban random bounces.

• Celebrate one moment. Pick a single action (Publish/Ship/Send). Make it feel earned.

Here’s my goal:

build Penfriend like it’s a luxury good - not precious, not performative, just inevitably better. Aesthetic that vanishes, performance that never makes you wait, experience that nudges you forward.

My worry? The market keeps rewarding cheap “AI-powered” theatre. Finance will side-eye anything that looks like craft. And I’ll be tempted to ship safe instead of special.

I’m going to try anyway. If it works, you’ll feel it in the first 60 seconds: faster first wins, fewer clicks, and the sense the app is on your side. If it doesn’t, I’ll show you where it hurt and what we ripped out.

Hold me to it—reply with the one flow you’d pay extra for if it just felt expensive to use. That’s where we start.

See you tomorrow.

✌️ Tim "What’s a medium luxury?" Hanson
CMO @Penfriend.ai

Same brain, different platforms: X, Threads, LinkedIn.

P.S. Asking for a me, do you think a luxury software is a thing?

Penfriend.ai
Made by content marketers. Used by better ones.