Independent AI evaluation of every SVG brand mark from Rounds 02-06. No knowledge of which symbols were manually selected as finalists. Two-pass pipeline: structural triage, then deep semantic scoring.
Phase 02 generated 850 SVG brand mark concepts across 6 rounds of exploration — from broad territory mapping (270 concepts across 9 directions) through focused refinement (110 variants of 11 shortlisted families). This page presents the results of an independent AI evaluation of all scorable SVGs, conducted with zero knowledge of which symbols were manually selected as finalists. The AI received only the raw SVG files and the Kurnik brand brief.
All 760 SVGs scored on 3 structural criteria (1–5 each, max 15). This pass evaluates the SVG as a graphic object — no brand context, no narrative, just shape quality. Top 76 advance (score ≥13, or 12 with a standout 5 in any criterion).
| Criterion | Evaluates |
|---|---|
| Geometric Clarity | Clean element count, well-defined shape relationships, no SVG artifacts or unnecessary complexity. Fewer elements with clear hierarchy scores higher. |
| Silhouette Strength | Does the mark have a distinct outline when reduced to a single-color blob? Good positive/negative space balance, effective use of the viewbox. Marks that "read" as a shape from across a room. |
| Uniqueness Signal | First-pass novelty check. Avoids immediately recognizable cliches (basic arrows, simple circles, generic grids). Novel compositions or unusual element combinations score higher. |
76 survivors scored on 6 weighted criteria (max 100pt). This pass evaluates the mark as a brand asset — how well it serves Kurnik specifically. Each criterion is scored 1–10, then multiplied by its weight. The weighted formula: (SC×2 + BN×2 + DI×1.5 + LH×1.5 + IP×1.5 + ER×1.5) ÷ 2.
| Criterion | Weight | Max | What it measures |
|---|---|---|---|
| Scalability | ×2 | 20 | Legibility at 80px, 32px, and 16px favicon. Stroke weights that survive reduction. Bold forms over fine detail. |
| Brand Narrative | ×2 | 20 | Does the shape tell a story about Kurnik? Progress, direction, focus, building, launching. Geometry with meaning vs. pure decoration. |
| Distinctiveness | ×1.5 | 15 | Visual territory ownership. Would this be confused with Stripe, Figma, Vercel, or any other tech mark? Novel in the landscape. |
| Lockup Harmony | ×1.5 | 15 | How well does the mark sit next to "kurnik" as a wordmark? Weight balance, alignment, proportions, breathing room. |
| Icon Presence | ×1.5 | 15 | App icon quality inside squircle/circle frames. Frame fill, visual mass, standout potential in an icon grid. |
| Emotional Resonance | ×1.5 | 15 | "Warm and confident" — infrastructure-grade quality with human warmth. Not cold, not playful, not generic. The Kurnik feeling. |
Each mark also receives a strengths and weaknesses text analysis explaining the score rationale. Click any mark to see the full breakdown.
Highest-scoring marks from the deep semantic analysis. Manual finalists are highlighted for comparison.
Where the 9 manually-curated finalists landed in AI rankings — and what the AI picked instead.
Score progression across rounds for each concept family. Families with Pass 2 entries shown.
Every mark scored, sortable and filterable. Cards with amber left border have deep (Pass 2) scores.