Slop*Detector
Verdict for demo.tikuma.net
SUSPICIOUSLY VIBEY — slop score 74/100
The hara architecture site piles up enough tells that the 74/100 score feels right, and I'd call this one with confidence. The em dash in the Villa Ruang Cahaya listing is the clearest fingerprint, but it's the cumulative pattern that seals it: every section opens with the same small uppercase eyebrow label, a '1.0'-style numeral that doesn't correspond to any real sequence, and a headline-plus-blurb rhythm that repeats almost identically from the ethos section down to contact. The copy leans on the same crutches too, from the rule-of-three in 'vision, spatial honesty, and craft' to the hollow reframe 'Great buildings rarely start with a drawing. They start with...' to a closing line, 'There is always more possible than you think,' that sounds weighty and says nothing. Two project cards show flat tan placeholder rectangles where photos should be, sitting right next to finished images, which points to unfinished generated content rather than a deliberate gap. None of this rules out a human hand somewhere in the process, but the gradient backgrounds, glassmorphism panels, and stock '42+ Commissions Built' stat row suggest a template built fast and dressed up with AI-flavored copy rather than a studio site shaped section by section.
Findings
- Feature-tile grid — Equal-width card grids of 3–4 identical tiles (icon, subheader, blurb) are the signature layout block of generated landing pages.
- Gradient backgrounds / gradient text — Large gradient washes and gradient-clipped headline text (usually blue→purple) are the default decoration of generated UI.
- Glassmorphism — Frosted-glass backdrop-blur panels are a recurring generated-UI decoration.
- Numbered X.0 section markers used as filler — Each major section is preceded by a large '1.0', '2.0', '3.0', '4.0', '5.0' numeral, but these aren't an actual ordered process or steps — they're decorative section counters applied to unrelated content types (ethos, disciplines, projects, team, contact).
- Tiny all-caps mini headers above section text — Nearly every section opens with a small uppercase eyebrow label with a leading dot, a pattern the guide identifies as a default AI section-header device repeated mechanically.
- Repeated kicker + headline + blurb hierarchy — The same three-level rhythm (small label, oversized statement headline, short explanatory paragraph) repeats almost identically across every section of the page, creating the generated cadence the guide warns about even without literal feature-grid cards.
- Placeholder/unfinished project images — Two project cards in the grid ('Studio & Roastery Kebun' and 'Rumah Halaman Hening') show flat blank tan rectangles instead of actual photography, while sibling cards above have real images — indicating unfinished or auto-generated content slots rather than a deliberate design choice.
- Placeholder image block in content section — A plain tan rectangle sits beside a real photo in the 'Creative thinkers. Disciplined builders.' section, functioning as an unfilled image placeholder rather than finished content.
- Faux-insight / manufactured-profundity copy — Multiple headlines use the generic profound-statement pattern ('Great buildings rarely start with a drawing. They start with...', 'There is always more possible than you think') that asserts depth without specific content, a tell from the writing guide applied to hero/section copy.
- Stat-block banner with suspiciously round/generic numbers — The hero stat row ('42+ Commissions Built', '100% BIM Integrated', '14 yrs Practice Lifespan') reads as a stock credibility-stat pattern common in generated landing pages, placed directly in the hero without further substantiation in the body.
- Em dashes — Em dashes are the single most reliable punctuation tell of LLM copy. Human marketing copy almost never uses them; models reach for them constantly.
- AI vocabulary — Words statistically overrepresented in generated text: "seamless", "leverage", "unlock", "empower", "elevate", "robust" and friends.
- Compulsive rule-of-three — The copy stacks triads repeatedly across nearly every section, a classic generative rhythm rather than natural variation in list length.
- "Whether you're X or Y" catch-all sweep — This is the textbook audience-sweep construction the style guide flags, used here to gesture at breadth instead of naming a specific client.
- "From X to Y" sweeping range — A generic range standing in for concrete description of the actual process, a hallmark of filler specificity.
- Disguised "X, not Y" reframe — "Rarely start with a drawing. They start with..." is the reframe trick in miniature: set up an expectation just to replace it, rather than stating the point directly.
- Manufactured profundity / hollow closer — A vague, aphoristic line that sounds weighty but says nothing concrete, used as a section-closing flourish typical of generated copy.
- Rhetorical-question opener — Opening with a question the following sentence immediately answers is paragraph machinery rather than a real address to the reader.
- Clipped parallel fragments — Two short declarative fragments stacked for rhythm rather than flowing prose, matching the banned drumbeat pattern.
- Generic cliché / empty claim — "Speak for themselves" is a stock phrase asserting quality without showing it, common filler in generated portfolio copy.
- Vague conceptual nouns repeated — Words like "poise" and "clarity" are used as abstract signifiers of quality rather than concrete description, and "poise" recurs suspiciously across unrelated sentences.