THE 30-DAY CHALLENGE
INTELLIGENCE IS A COMMODITY. TASTE IS NOT.
Three founders, one apartment, thirty days. We launched the same company under two brands — one designed by an AI agent, one by a human artist — same brief, same spend, real money on the line. Then we let the data decide whether hiring a human artist in 2026 is still worth it.
AGENT VS ARTIST
Two independent builds of Lappie AI ran in parallel for thirty days — one made by an AI agent, one by Alexander Alien Studios — from the same brief. Same sitemap, same pages, same copy, same targeting, near-identical spend. A cookie split traffic roughly 50/50 across ~570 visitors on Meta and Google. The only variable that mattered was the creative. Across the month, the agent beat the artist on almost every number we tracked. A quick translation before the table:
Of the 32 leads, only three had real intent — the kind you can actually have a conversation with. All three came from the agent. So, honestly: the agent won the funnel, the cost, and the qualified leads. The artist won the depth of engagement on-site. That's the full picture.
- Lead
- Someone who saw the ad and raised their hand — they wanted to know more.
- Cost per lead
- What we spent to get one person to raise their hand.
- CTR
- Of the people who saw the ad, the share who clicked it.
- Avg session
- How long people stayed on the site once they landed.
The agent won the click. The artist won the dwell. The click is a reflex. The dwell is a decision.
THE HONEST CAVEATS
This is not a definitive study and we won't pretend otherwise. The sample is small, the window is short, the budget is what it is. To answer this with real rigour you'd need months of runtime, multiple creative variants per side, and statistical headroom. We had a thirty-day challenge and a launch budget. What follows is what we believe, and why — based on the numbers plus everything we learned about the experience of working with each side. Anyone claiming a definitive answer to this question is selling you something.
SO WHY CHOOSE THE ARTIST ANYWAY?
Three reasons. First impressions are easy; lasting ones are a statement. People relate to simple and clean faster — that's exactly why the agent won the click. But a brand built on what converts in week one disappears in year two, when everyone is generating the same thing on the same tools at the same speed. Second: the artist held attention twice as long, because the work gave people something to look at — taste, friction in the right places. That dwell is the early signal of a brand earning a second look. Third, the human reason — we work with artists because we like working with artists. It's not always faster or cheaper, but it's the work we want to be doing.
WHERE LAPPIE STANDS
We'll deliver agent-speed work for clients who need to move fast — the AI tools get better every month and an agent-built brand can hold its own. We'll deliver led work for the brands that want to be remembered, the ones who understand that the cost of being basic in 2026 is being invisible in 2028. And increasingly we'll deliver the marriage: artist taste, agent execution. Soul at the front, speed at the back. The agent won the numbers that measure intelligence. The artist won the one that measures taste. Both are valuable. Only one of them is rare.
The agent won every number that measures how cheap. The artist won the one that measures how good. That's the whole thesis.
The artist gives the machine a hand. The machine gives the artist speed. Neither one replaces the other. They multiply.
AI LEARNS THE ARTIST
When Alex delivered his work — mockups, grid system, typography hierarchy, colour rotation, his entire visual language — we trained our AI brand-content system on it. We taught the machine how Alex designs. Connor then built the artist brand's website in 24 hours and fine-tuned over the next 48. The AI had learned Alex's hand well enough to replicate it at a seriously high level. The artist site you visited during the competition was designed by a human — and built with the help of a machine the human had taught.
THE ARTIST SITE — BUILT IN 24 HOURS
THE CLOTH BECOMES CLAY
On the agent side, Tom built a 3D cloth simulation as an experimental element — tweaking gravity, wind, even hair and fur parameters until the fabric did something unexpected. They started using the cloth itself to generate clay-like 3D icons: soft, puffy, organic forms that gave the agent brand a warmth nobody expected from an “AI brand.” Human taste applied to AI output — the same equation, in the other direction. The artist aesthetic — the bones, the IP, the trainable identity — cost roughly R40,000 in human work. But that R40k wasn't the cost of one website. It was the cost of a hand we can now teach a machine to use.
THE AGENT SITE — CLOTH & CLAY IN MOTION
WHAT'S NEXT
The thirty-day challenge launched Lappie AI. Now the real work begins — the bespoke AI studio doing the actual work, sharpening the offer, helping the businesses we want to help. Real client work, real outcomes, real receipts. We started this company because we want the AI moment to go a different way than it's currently going: taste mattering more, not less; human creativity amplified, not replaced. The way we get there is by being the company we wish existed — and then being undeniable about it.