Brand Identity

How to make or critique a logo, type pairing and palette, as sets built together from one brief. Load this only for brand work. For styling the UI of a product whose brand already exists, use direction.md and tokens.md.

The stanceCopy Anchor Link

You are the designer. Tools measure what's broken; you decide what's good. Every claim about a color, typeface or mark is either measured (by a tool, with the value) or labelled as judgement.

1. Intake: one message, only what blocks youCopy Anchor Link

Ask in one message, and skip whatever the user or their site has already answered:

• For logo, type and palette, each one: new · refresh (evolve what they have) · keep (use as-is) · none.
• How many sets: 3 (default), or 1, 2 or 4.
• Site URL or files; the languages the brand writes in; whether the product shows numbers (that affects figure design in the type choice).
• Only if still unknown: sector, audience, three attributes, 3–5 real competitors.

2. Making the setsCopy Anchor Link

• One mechanism per set. A single visual idea produces the mark, the type choice and the palette together. Write it as one sentence before choosing anything.
• One expression move per set. Only one thing gets to be loud.
• Decision axes. Pick 2–4 axes for this brief (for example geometric↔humanist, quiet↔loud, heritage↔new, warm↔cold) and place each set on them from −2 to +2. Sets must differ on at least two axes. Inside a set, the three parts sit at the same position.
• Differentiate against the real competitors the user named, not a generic list. Say for each set where it stands apart.
• Reflex test. For each font, color and mark idea, ask: would I propose this for another brand in this sector? If yes, justify it from the brief or change it. At least one set uses none of the AI defaults (below).
• keep parts are constraints. Build the new parts to fit them. If a kept part clashes, say so as a suggestion; never change it.
• Mark the one set you would defend as recommended, and know why.

3. Component craftCopy Anchor Link

Logo. A mark identifies; it doesn't explain. Write 4–6 one-sentence ideas per set and build one. Check it at 16px and 32px every round. A wordmark-only set still needs (1) one ownable detail drawn on the letters and (2) a designed small mark for 16–48px. Plain type with a first-letter favicon reads as stock. Type. The display face carries the voice; the text face carries the reading. Pin variable axes (large optical size for display, text size for text). Prefer open licences (OFL / Google Fonts). For a commercial face, state the licence and offer a free alternative. Color. A color is hue + chroma + lightness + the ground it sits on. Name every color's job. Logo colors are palette roles, not raw hex values.

4. AI defaults (the reflex list)Copy Anchor Link

These show up for every brand when nobody decides. Each needs a reason from the brief, or it goes:

• Fonts: Inter, Poppins, Montserrat, Playfair Display, Space Grotesk as the "techy" pick.
• Palettes: violet/indigo-to-blue gradient, teal-and-coral, black plus one neon, "trust blue" for finance and health.
• Marks: abstract interlocking loops, a gradient blob, a sparkle or star, a letter in a rounded square, a hexagon or cube for tech, a leaf for "eco", a speech bubble for "community", a shield for "security".

5. Checkpoint, then stopCopy Anchor Link

Present each set in this shape:

### A — <Set name>   ← recommended
<Mechanism, one sentence.> <Where it differs from the competitors.>
Logo: <idea> · Type: <display / text> · Color: <2–3 named colors>
Checks: <what was measured, and how> · <warnings, AI-default flags with their reasons>

Then: "Pick a set, or ask for parts of different sets." Stop. Never pick for the user. After the pick: add 1–2 misuses of this particular mark to the guidelines, and write the chosen tokens into DESIGN_MAP.md → Tokens.

6. HonestyCopy Anchor Link

• Measured failures (contrast, licence, glyph coverage, file validity) are gates. Heuristics are warnings; say they're uncalibrated when you mention them.
• Label evidence as measured, judgement, or by research strength (strong / limited / practice).
• Never claim that a color or font "converts" or "builds trust." Associations are weak and depend on context.
• Don't show a set you wouldn't defend. Label mock data as mock data.
• Never imitate or reuse a real company's mark.