Google Maps Lead AI

Turn any Google Maps search into a pre-qualified lead list — with real phone numbers and AI classification.

Scrape businesses from Google Maps — name, phone, website, address, category, rating, review count, review excerpts and coordinates — then let AI sort every result into your own labels and tags. You don't just get a contact list; you get one that's already triaged by fit, specialty and reputation, ready for cold-calling, prospecting, CRM enrichment or market research.

---

Why this scraper

• 📞 Real phone numbers — pulled from each place's detail panel, where Google actually exposes them. Many cheaper scrapers skip this; the phone is the whole point of a lead.
• 🧠 AI lead classification, your way — bring your own labels and/or describe in plain language how to classify. Get back structured, grouped tags for every lead.
• 🔁 Incremental mode for scheduling — on repeat runs, only new places are visited and charged. Perfect for a daily or weekly cron.
• 🌎 Any niche, any country — "lawyer Belo Horizonte", "dentist Austin", "real estate agency Lisbon". You define the searches and the labels.
• ♻️ Deduplicated & cached — no duplicate places; AI classifications are cached for 7 days, so re-runs never re-pay for the same place.

---

What you get — per place

{
  "name": "Silva & Associados Advogados",
  "phone": "+553133334444",
  "website": "https://silvaassociados.com.br",
  "category": "Advogado",
  "address": "Av. Afonso Pena, 1000 - Centro, Belo Horizonte - MG",
  "rating": 4.8,
  "reviewsCount": 127,
  "reviewsSample": ["Excellent service, highly recommend..."],
  "lat": -19.9245,
  "lng": -43.9352,
  "labels": ["LAW_FIRM"],
  "tags": { "area": ["labor", "civil"], "reputation": ["good reviews"] },
  "confidence": 0.92
}

Every run streams clean, deduplicated JSON to the dataset — export to JSON, CSV or Excel, or pull it straight from the API.

---

AI classification — two knobs, one cheap call

Classification is optional and uses Groq (bring your own API key from console.groq.com). Use either knob, or both — everything runs in a single AI call per place, so extra characteristics cost almost nothing.

1. Labels — a strict shortlist

Provide a closed list (e.g. LAW_FIRM, SOLO_LAWYER, NOT_A_FIT). The AI assigns the 1–3 that fit, returned in labels. Ideal for filtering out the noise Google mixes into a category — accountants, notaries and "consultancies" that aren't your target.

2. Instruction — free text, grouped tags

Describe, in your own words, exactly how you want each place classified:

"Classify law firms by practice area (labor, civil, family, criminal) and by review reputation (good reviews / complaints), with separated tags."

→

"tags": { "area": ["labor", "family"], "reputation": ["good reviews"] }

The AI groups its findings by each characteristic you asked for, and returns a confidence score (0–1) with every result.

---

Incremental runs — built for scheduling

Turn on Only New Places and the actor remembers every place it has collected. On the next run it reads the search results, skips the expensive detail visit for places it already has, and processes only the new ones.

A scheduled daily or weekly run then surfaces only fresh leads — instead of re-scraping, and re-paying for, the same city every time. Leave it off for a one-shot full pull.

---

Input example

{
  "searchTerms": ["advogado"],
  "locations": ["Belo Horizonte, MG"],
  "maxPlacesPerSearch": 100,
  "classifyWithAI": true,
  "groqApiKey": "gsk_...",
  "labels": ["ESCRITORIO_ADVOCACIA", "ADVOGADO_SOLO", "NAO_FIT"],
  "classificationInstruction": "Classify by practice area and review reputation, with separated tags.",
  "onlyNewPlaces": false
}

Only searchTerms and locations are required. Each search term is combined with each location.

---

Great for

• Lead generation & cold calling — a phone list that's already qualified by fit, specialty and reputation.
• Sales prospecting at scale — schedule it and receive only new leads on every run.
• Market research — map who's in a category, where they are, and how they're rated.
• CRM enrichment — feed structured, tagged businesses straight into your pipeline.

---

Good to know

• Phone numbers live on each place's detail panel, so the actor visits every place — accurate contact data, powered by a real browser and residential proxy.
• AI classification is optional (bring your own Groq API key). Leave it off for a pure scrape.
• Deduplication is automatic (by place and by name), so your dataset stays clean across searches and runs.