Namer

Namer

Structured naming pipeline for any product, company, or brand. Uses the Diamond Framework (David Placek / Lexicon Branding) for strategy, then leverages LLM knowledge of languages, cultures, history, and sound symbolism to generate and evaluate candidates.

Pipeline: Discover → Seeds → Generate → Evaluate → Verify → Present

Step 1: Discover

Ask these 5 questions one at a time, in order:

1. What's the domain? — "What industry or category is this?" (e.g. coffee shop, developer tool, law firm, band)

2. What does winning look like? — The vision. What is this, who's it for, what does success look like? > Swiffer: "Build a mop-like device people pay a premium for. Make cleaning floors something people want to do."

3. What do we have to win? — The advantage. What assets or insights give you an edge? > Swiffer: "P&G's Pampers diaper tech — a lighter, more effective tool using absorbent pads instead of water."

4. What do we need to win? — The gaps. What's missing? What must be overcome? > Swiffer: "Avoid being seen as just another mop. People hate mopping — they need to see this as entirely new."

5. What do we need to say? — The message. What should the name communicate? What should it feel like? > Swiffer: "Logical: efficient, quick, easy. Emotional: fun, joyful, light. Should sound like a quick, satisfying action."

See references/diamond-framework.md for the full framework with case studies.

Step 2: Extract Seeds

From the user's answers, extract 30-50 seed words:

Literal seeds (~15-25):

• ~3-5 from domain vocabulary
• ~3-5 from "winning" (vision/ambition)
• ~3-5 from "have to win" (differentiators)
• ~2-3 from "need to win" (aspirations)
• ~3-5 from "need to say" (feeling/tone)

Evocative seeds (~15-25) — generate these yourself:

• Metaphorical — words from adjacent domains (nature, mythology, architecture, music, materials). "If this product were a natural phenomenon, what would it be?"
• Cross-cultural roots — Latin, Greek, Sanskrit, Japanese, Arabic roots carrying the right connotation
• Sound-designed — invented syllable combos chosen for phonetic personality

Sound profile — from "need to say," define the target phonetic character using references/sound-symbolism.md. Identify target phoneme classes (plosives, fricatives, nasals, etc.), vowel types (front/back/open), and specific high-value letters.

Present the seeds and sound profile to the user before proceeding.

Step 3: Generate

Generate 150-200 name candidates using these approaches and references/morphemes.md as a source:

1. Cross-cultural roots — Latin, Greek, Sanskrit, Japanese, Arabic roots from references/morphemes.md related to seeds. Remix, combine, truncate into name-length forms.
2. Mythological & historical — figures, places, concepts from world mythology and history that resonate with the brand strategy.
3. Morpheme blends & portmanteaus — combine syllable fragments from seeds into invented words. Target 5-9 characters.
4. Compound multipliers — two short words/fragments combined where 1+1=3 (creates richer associations than either alone). E.g. BlackBerry, Impossible.
5. Sound-symbolic inventions — coinages built from phonemes matching the target sound profile. Not derived from real words — pure phonetic design.
6. Industry vocabulary remixes — take domain-specific terms and transform them (truncation, blending, respelling, suffix swaps) into ownable forms.
7. Unexpected assemblies — familiar word parts combined in unfamiliar ways. Palindromes, letter pattern plays, respellings.

Generation rules:

• Target length: 4-10 characters (sweet spot: 5-8)
• Must be pronounceable in English
• Must be spellable from hearing it spoken
• No existing common English words (they're unownable in search/SEO)
• Separate creation from judgment — generate all candidates before evaluating any
• Aim for diversity across approaches, not all from one category

Step 4: Evaluate

Filter to top 20 candidates (ranked) using these criteria:

Placek's Three Pillars:

1. Build for trust — name inspires confidence. Familiar components assembled unexpectedly (Azure: "z" = signal, "sure" = confidence). Invented names that feel like real words.
2. Communicate an original idea, never describe — descriptive names are boring and have thousands of competitors. The name should make people say "they're not like the other guys."
3. Be accessible — easy to process, pronounce, spell, remember. The brain doesn't like complexity.

Additional filters:

• Strategic alignment — does it serve the Vision, Advantage, Gaps, and Message from discovery?
• Sound symbolism match — does the phonetic character match the target sound profile?
• Compound multiplier — if a compound, does 1+1=3? Does it create richer associations?
• Competitor gut check — "Your competitor just launched with this name. Reaction?" If you'd shrug, it's too weak.
• Discomfort signal — if the name feels safe and comfortable, it's probably wrong. Great names make you uncomfortable at first.

Keep positions 11-20 as a bench for replacement if top names get bumped in verification.

Step 5: Verify Availability

Web-search each of the top 20 candidates. Two searches per name:

1. Bare name: "[name]" — is it already a brand/product?
2. Domain-specific: "[name] [domain from Step 1]" — is it taken in this space?

Decision per name:

• Nothing relevant → CLEAR — stays
• Hits exist but not in same space → CAUTION — stays, note the conflict
• Existing product/company in same space → BUMP — remove, pull next from bench
• Common dictionary word with millions of results → BUMP — unownable in SEO

Keep going until you have 10 CLEAR/CAUTION names.

Optional extras (suggest to user, don't assume):

• Domain availability (.com, .io)
• Trademark search (USPTO)
• Social handles (@name on major platforms)
• Software registries (npm, PyPI, crates.io, GitHub)

Step 6: Present Top 10

For each finalist:

• The name
• Why it works — tied to Diamond Framework strategy
• Sound notes — phonetic character and symbolism
• Cautions — any availability concerns
• Domain — if checked

Key Principles

• Invented names take LESS money to build than existing words. They signal "new and innovative" and generate more attention. Humans are drawn to the new.
• Great names don't describe reality — they change it. Swiffer made mopping feel like a quick, joyful action.
• Quantity breeds quality — but intentional quantity, not mechanical volume. Every candidate should have a reason to exist.
• Compounds outperform single words for memorability.
• Don't brainstorm with 4+ people — peer pressure kills originality.
• The domain question changes everything — "Atlas" is clear for a coffee shop, taken for software.

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: jbold
• Source: jbold/namer
• License: MIT

Install and usage instructions live in the source repository linked above.