Customer Lookalike Finder
ethan-ab/customer-lookalike-finder-desktop
Encontre prospects B2B similares aos seus melhores clientes direto no Claude
Ver no GitHubPlugin para Claude Desktop que identifica empresas B2B similares aos seus melhores clientes, sem API keys ou dependências Python. Filtra automaticamente mega-corporações e sites de reviews, retornando uma lista ranqueada de prospects com estágio e tamanho compatíveis.
Instalação
npx convez-ai ethan-ab/customer-lookalike-finder-desktop1 comando no terminal: detecta o formato do repositório (marketplace ou plugin avulso) e instala via Claude Code. Requer Node 18+ e o Claude Code instalado.
Métricas
Casos de uso
- Prospecting qualificado para SDRs e equipes de vendas
- Identificar concorrentes e empresas de perfil similar para análise competitiva
- Descobrir parceiros potenciais no mesmo segmento de mercado
Veredito

Plugin prático e bem-arquitetado para prospecção B2B, com filtros inteligentes contra alucinações e mega-corporações. Funciona apenas com web search nativo do Claude, o que limita a profundidade de análise, mas mantém a simplicidade e não requer configuração.
README
A tradução para o português deste README ainda está na fila.
customer-lookalike-finder
Find B2B companies that look like your best customers — right inside Claude. No API keys, no Python, no setup.
Give it the domain of one of your best customers. It returns a ranked list of similar companies that match that customer's size and maturity — automatically filtering out mega-corporations, listicle/review sites, and anything it can't verify.
Built for SDRs, founders, and GTM teams who want a fast, defensible lookalike list without leaving the chat.
Install
/plugin marketplace add ethan-ab/customer-lookalike-finder-desktop
/plugin install customer-lookalike-finder-desktop
Or import the customer-lookalike-finder-desktop/ folder as a local plugin in any Claude client that supports plugins. It needs only web search and fetch — no API keys, no Python.
Usage
/lookalike legora.com
Multiple seeds at once:
/lookalike stripe.com adyen.com checkout.com
If you omit a domain, the plugin asks for one.
How it works
Claude runs the full pipeline itself, using the web tools available in your session:
- Understand the seed — fetches the homepage and infers category, target buyer, and stage.
- Generate candidates — web searches for alternatives, similar companies, and category peers (15–25 raw candidates per seed).
- Filter — drops mega-corporations (>5,000 employees), listicle & review sites (G2, Capterra, Crunchbase, …), and any company it can't verify exists.
- Score (0.0–1.0) — category match, buyer match, and stage match, with penalties for scale mismatch and non-SaaS business models.
- Output — a ranked markdown table.
Example
/lookalike legora.com (a legal-AI scaleup) returns:
| score | domain | company | why_match |
|---|---|---|---|
| 0.92 | harvey.ai | Harvey | Legal AI for law firms — review, research, draft |
| 0.88 | spellbook.legal | Spellbook | AI drafting/review for transactional lawyers |
| 0.85 | robinai.com | Robin AI | Contract AI for legal teams |
| 0.83 | luminance.com | Luminance | AI document review for legal departments |
| 0.82 | legalfly.com | LEGALFLY | In-house legal AI — review, draft, compliance |
| … |
Mega-corps such as Thomson Reuters' CoCounsel and LexisNexis are filtered out automatically, even though search engines surface them as "alternatives."
Design principles
- Maturity-matched — a 50-person Series A seed will not return Microsoft.
- No invented companies — better 5 real lookalikes than 15 hallucinated ones.
- Honest reporting — if filtering leaves few results, it says so instead of padding the list.
Limitations
- Results depend on the quality of Claude's web search; they are not citation-locked.
- Works best for B2B SaaS / tech seeds with a clear product category.
License
MIT — see LICENSE.