Nights & Weekends

Before there was a name, there was a question.

I was running a pre-buildspace school over Zoom in 2021. Six weeks. I taught a small group of people the basics — design, marketing, writing — then asked each one the same question at the end:

What are you working on when no one's looking? What are you doing on your nights and weekends?

The answers were good. Better than the curriculum. The thing they made when no one assigned it was always more honest than anything I'd asked them to make.

That question became the brief for the next thing.

The thesis

Most people's best ideas don't happen at work. They happen at night. On the weekend. In the gap between the thing they're supposed to be doing and the thing they can't stop thinking about. The hustle-culture version romanticizes the side project as proof of grind. We did the opposite. We made the side project the main thing.

Nights & Weekends was a six-week program for builders already deep in something — not for people thinking about starting. Not for people planning to start. For people who'd already been at it for a while, alone, and didn't know there were others like them.

We ran six of them. The growth tells the story better than I can:

• Season 3 — 7,000 builders
• Season 4 — 15,000 builders
• Season 5 — 70,000 builders

I led design across the whole arc — the brand, the houses, the per-season systems, the unicorn, the acceptance assets, the demo-day visuals, the operating system the builders lived in. What follows is the design story underneath those numbers.

The unicorn

Every program I'd ever seen used a logo that looked like a program. A wordmark. A cohort number. Clean, institutional, forgettable.

Nights & Weekends needed a symbol, not a logo. Something a first-time builder could put on a sticker, a hoodie, a wallpaper, a tweet — and want to. Something that didn't feel like school.

The unicorn worked because it broke the rule. Tech doesn't usually allow itself a mythic creature. We allowed it. The unicorn became shorthand for the kind of builder the program was for — rare, weird, hard to find, often unbelievable until you saw the thing they'd made.

That's the trick of a good mark. It does the sorting for you.

The four houses

By season four the program was too big to be one thing. Fifteen thousand builders need sub-cultures, not just a Discord. They need belonging at a smaller scale than the cohort.

So we built four.

• Spectreseek — methodical, technical, head-down. Jeff energy.
• Gaudmire — bold, fast, loud. Farza energy.
• Alterok — the explorers. Off-the-map builders.
• Erevald — the dreamers. World-builders.

Each house got its own identity — name, wordmark, color, vibes video, wallpaper, acceptance pack, website (spectreseek.com, alterok.com, gaudmire.com, erevald.com), Discord channel. Each had a sorting quiz. Each had house leaderboards inside the program leaderboard.

What surprised me is how fast it worked. Within a week of houses going live, people stopped saying I'm in nights and weekends and started saying I'm in Gaudmire. The smaller identity was stickier than the larger one. They invited friends into their house. They started inside jokes. They argued about which house was the best. The thing I'd designed as a UI feature turned into a cultural one.

If I were doing it again I'd push this further. Houses are a way to make a 70,000-person community feel like a 200-person one without splitting the room.

The per-season system

Each season was six weeks. Tuesday lectures. Thursday workshops. Weekly updates. Demo day at the end.

The shape stayed the same. The skin changed every season — new acceptance pack, new wallpapers, new slide templates, new title cards. Familiar enough that returning builders knew the rhythm, fresh enough that it didn't feel like a rerun.

The lecture system was a design problem more than a content one. We were producing two pieces of video a week, for six weeks, at a quality where 70,000 people would actually watch. The fix was a template. Every lecture deck was the same skeleton — title card, hook, beat, demo, ask — with the seasonal skin layered on top. The system did the work. We could spend our time on the lecture, not on the slides.

The smaller surfaces did more work than I expected. The chrome extension turned the new-tab page into a builder dashboard — a countdown to demo day, the week's prompts, the leaderboard. The emails went out in lowercase, written like a friend telling you something, not a brand asking for attention. The 90-second final pitch went out to the whole mailing list. None of these were the headline asset. All of them were the program.

The other thing I'd flag: the weekly update was the engine. Every Sunday, every builder posted (a) their toy, (b) a video of it, (c) feedback to five other people's toys. That's it. Three things. Public. Repeatable. It's the simplest design in the whole program and it's the one that made the program work.

What it felt like from the inside

I had the asset side. The user side I only got secondhand — mostly through writeups builders posted after each season. The pattern in those writeups did more for me than analytics did.

People remembered the chrome extension. They remembered the houses by name. They remembered that the emails didn't sound like a brand. Cam Houser, who made the top 16 of season three, called the program "heavy on vibes" — the politest version of the thing we were going for.

The clearest signal was a small one: builders didn't describe nights & weekends the way we did. They described it as a place. Season five ended with a thousand of them flying to Fort Mason for three days with no agenda. People paid airfare to attend an event we'd deliberately under-programmed. That's the metric I cared about — whether someone would fly across the country to stand next to other people who'd been inside it with them.

What I'd do differently

Three things, in order:

1. Make the houses earlier. We didn't add them until season four. They should have shipped with season one. The sorting + the sub-cultures are most of what people remember.
2. Write down the lore. The unicorn, the houses, the rituals — they all happened, but most of the story lived in heads. The next version of this needs a canonical document so it survives the people who built it.
3. Take more of the receipts. I have the assets. I don't have enough of the wild — the screenshots of people putting the unicorn in their bio, the houses on a hoodie, a season slide template ending up in someone else's deck. The proof of belonging is in those, not in the assets themselves.