ACC Business Scraper

The Arizona Corporation Commission publishes every new LLC filing in the state. That's a live feed of people who just started a business and don't yet have a website, an accountant, or a real operation.

It's a gold mine for cold outreach if you can get to the data first.

In January 2026 the ACC replaced its old eCorp portal with a new system at arizonabusinesscenter.azcc.gov. The old scrapers in the wild stopped working. I built a new one.

No public API. Everything is rendered client-side and requires interaction to reach the filings list.

Rate limiting on the portal punishes aggressive scraping.

Filings include LLCs that are formed for legal structure but have no operating business behind them. Those are noise.

Contact data is not in the filings. You get a business name, an agent, and an address.

Built a Playwright scraper that runs on a schedule, hits the portal during low-traffic windows, pulls new Articles of Organization filed in Maricopa County within the last 24 hours, and deduplicates against a local seen-ids.json.

Enrichment pipeline runs after scraping. Each new business goes to Perplexity with a prompt designed to find a website, a phone number, and an owner name. Perplexity's web grounding beats every structured data vendor I tested for brand-new businesses that don't have a web presence yet.

Scored leads and custom call openers get generated by Claude, pulling from the business name and any enrichment context. Output is a CSV ready to drop into a dialer or into cold email sequencer.

Playwright · Node.js · TypeScript · Perplexity API · Anthropic SDK · Docker · Hetzner VPS · cron

Self-sustaining lead pipeline that feeds new Arizona businesses into cold outreach within 24 hours of them existing on paper. Runs unattended on a Hetzner VPS.