Ship Log

I launched Artemis Nova as my product consultancy last January, but I still didn’t have a website for the business. I had a vision in my mind for something simple, elegant and medieval inspired. My creative constraint was time: I had to ship it within a single work session. 4 hours later a website was born!

Problem I’m solving

While I was not actively taking on new clients at the time, my product consultancy needed a website to appear more legitimate, and a front door for new client requests. Who knows, maybe someone special would come along.

Where I started

My original vision was way grander - a gorgeous medieval style “book” you would open to see different custom generated images representing different projects in my portfolio. Think an illustrated manuscript like the Book of Kells. I also did some research into beautiful interactive portfolio sites like igloo.inc. However, I realized that to generate images in the style I wanted would be an entire intense sub project likely using image generation tools like Midjourney I’d never used before. Given my time constraints, I scoped the entire project down to a single page website that had the elements of my desired aesthetic.

Where I ended

I was pretty happy with the end result - an elegant single page website with a moon graphic that updates with the live phase of the moon. The moon is an important symbol to me personally, Artemis is the goddess of the moon, and I wanted the website to be feel slightly sacred/mysterious rather than corporate. I spent a lot of time on the moon image itself to get a “hand drawn” feel, and tuning the moon and logo “shimmer” to be less of a sweeping gradient effect and more of a dark gold glow effect.

What I learned

Design is very important to me and takes time, even for very simple experiences. In fact, the simpler an experience is, the more important design becomes. I always underestimate how much iteration it will take to get something to a place where I feel it is good enough, and it’s difficult to know when to cut it off. Having a time limit is really important because otherwise I can endlessly churn on improvements and iterations.

When I reviewed my work on this site, Claude actually said it best - getting good design from LLMs is more like sculpting. It takes a lot of patient micro-prompting.

At the beginning of the year, I had a goal to ship something every week. Life got busy and while I started with great intentions, I haven’t shipped anything for months because my focus has been on my hourly contracting roles and making $. My solve — a public log of things I make.

Problem I am solving

It is an effort to give myself a kick in the pants and some accountability. I hope to look back on it at the end of the year with a wistful sense of pride. With each thing I create, I learn something, use new tools, get a little sharper.

What I learned from this build

With a simple prompt, Claude is more likely to overbuild than underbuild. A lot of coming to something decent is stripping stuff down.

That being said, I always underestimate how long it will take to build a “simple” feature. Classic example — just a weekly log of what I shipped. But maybe also a visualization of the weeks… maybe the visualization could be animated… maybe the visualization could have a streak associated with it… maybe it should show what tools I used… maybe it should impact the dialog tree on the homepage. It goes on and on! But that is the fun of building products.

Also, confirming that I really dislike having Claude write for me. Despite building a custom skill that is supposed to help Claude learn to write in my voice, I just haven’t cracked the code on that. It is always too clever, too choppy, too something.

My family kept asking me about my trip to New Zealand so I built this micro site to show our itinerary to them.

Where I started

Built in frenzy during a 30 min on a layover before getting on the red eye to Auckland.

Where I ended

A pretty cool fully functional site showing current location, dispatches, date logic, etc.

What I learned

Don’t build a site that you have to maintain while on vacation! After I launched the site and sent it out, I had zero interest in pulling out my laptop to update it. So it kind of languished after that. But this is probably the fastest build I’ve ever done.

In the middle of the night, came across a few posts on Twitter about “AI psychosis”. It was an exhilarating time - agents had just gotten way better, and everyone was feverishly building stuff. I thought to myself that these posts didn’t quite capture the exhilaration and euphoria of the moment. It wasn’t psychosis, it wasn’t anxiety — it was mania! Thus “Cyclical Generative Mania” was born as a concept.

Where I started

Started as a joke Twitter thread I was going to post. Then I started to get deeper on the concept as my own CGM took hold…

Where I ended

I purchased the domain cyclicalgenerativemania.com and set up a full interactive website with a quiz, comorbid conditions, Promptzac™, etc. I tried to lean into the big pharma UX patterns as much as possible. I also developed an entire launch strategy on Reddit, X, Hackernews, etc. I was totally convinced it would go viral.

What I learned

It did not go viral lol. It was quite the opposite, I think it got about 2 likes from my tiny X account of 12 followers. Reddit didn’t like it, Hackernews didn’t like it, no one liked it but the friends in tech I sent it to, who said it was hilarious. My husband was like - why??? why would you spend time on this?? what about your actual startup?? All I could do was shrug because the whole thing was so ridiculous and meta at that point, a true manic sleepless Claude-fueled episode.

A weird thought in the middle of the night…why aren’t websites more fun? An image popped in my head - an old man by the campfire in the middle of the night. “Welcome weary traveler…I hear you’re looking for Ingrid.” Months later with Claude’s help, my weird personal website became a reality in a 2 hr ShipSLC build session!

Problem I’m solving

I always wanted a place to start to share my thoughts, ideas, projects. It also felt awkward as a PM that I basically zero web presence. For a person who spends their entire working life in the digital realm, it feels akin to being homeless or not having a front door to greet people at. At the same time, I didn’t want a typical portfolio page, I wanted something a bit more unique that captured my personality. Having played a lot of classic RPG games, an RPG style pixel art scene seemed like a great playful fit.

Where I started

With Claude’s help and some research, I decided to build on top of Astro, since I wanted a basic content/blog website. I decided to descope a fully AI-driven dialog to start, because I wanted to complete the build in 2 hrs and I actually thought the basic NPC-style dialog tree would be charming. I wrote a spec for the website, refined it with Claude, and thought I would be able to one shot it. Not quite - ha!

Where I ended

After a LOT of visual iteration, a beautiful pixel art homepage with a working dialog tree routing to the About section or the Blog. It was a really exhilarating fast build. The compromise I made was that my first blog page plus the About page were entirely AI generated content not written by me. In retrospect this was a bit weird if the site was meant to be a form of self expression, but overall I was really pleased with the end product.

What I learned

The before and after screenshots for this build are hilarious. The key difference between the two was reference images! I tried several frustrating rounds of trying to describe a better aesthetic to Claude with no progress. Finally went to Dribbble, searched for some beautiful pixel art, uploaded it, and the quality of the design went up 1000%. It was a massive unlock I’ll never forget. It still took about 10+ reprompts though to really dial in the imagery.

Also originally I didn’t have Playwright installed and Claude had no way to visually QA the website, which was a huge problem for something so design heavy.

Me not knowing the right words to use in my design tweaks was a problem. I wanted the homepage screen to be full screen and I didn’t realize the right word to use to describe this was “viewport”. Took a while to communicate what I wanted and have the full screen effect I was going for.

Meetup was going to cost $540/yr without pulling any new people in, was a total waste of money. I migrated to Luma.com to host my event calendar and never looked back!

Where I started

Had an old “Vibe Code Together” weekly event on Meetup. To be able to host a weekly event, Meetup forced me to upgrade to the Meetup Plus plan which was going to cost a fortune with very little value add. I decided it wasn’t worth it and Meetup DOES NOT LET YOU downgrade your plan after upgrading, so it was not possible to continue using Meetup. I found Luma after searching for alternatives and was really pleasantly surprised by how fully featured it was without all of the ridiculous paywalls and lockins. Also realized that “Vibe Code Together” was starting to feel outdated. Pretty sure “Vibe Code” will end up being a word of 2025 that won’t carry into 2026.

Where I ended

Migrated to Luma which ended up being amazingly easy to integrate with. Ended up with a new “Weekly Build” event format with different branding.

What I learned

Read the fine print. Meetup is predatory! I’m somewhat disgusted by their inability to downgrade an upgraded plan…very sleazy.