Business Recon

One-liner: A B2B lead discovery and enrichment pipeline — SMTP-verify emails, deep-crawl company websites, and render the results as a CRM view.

Origin (the Build Loop)

Because I needed to find and qualify B2B leads without paying for ZoomInfo or Lusha, I kept hitting the fact that every lead-generation tool is $100+/month and the underlying data is mostly public — you just have to find it on the right three pages, so I built a pipeline that SMTP-verifies email addresses, deep-crawls company sites, and presents the enrichment as a CRM view, that **takes a raw lead and returns a research file ready to send. It taught me that most business data is “public but not published” — it lives on an about-us page, a LinkedIn profile, or a PDF that Google crawled three years ago.

The problem

B2B lead generation tools (ZoomInfo, Lusha, Apollo) are expensive and their data is often stale. The raw information — company size, funding, tech stack, recent hires, email format — is public on the company’s own website, LinkedIn, and industry directories. The work is finding it, verifying it, and formatting it. I wanted to enter a company name and get back a complete lead file without paying per-record.

What it does

• SMTP email verification — checks whether an email address actually exists on the recipient’s mail server without sending mail (catch-all detection + mailbox verification).
• Deep website crawl — given a company domain, discovers about-us, team, careers, and press pages and extracts contact info, team size, and technology signals.
• CRM view — presents the enriched leads as a structured table with status, verification score, and enrichment depth.
• Retry logic for unknown SMTP verdicts — servers that return “try again later” get requeued, not dropped.
• Admin actions — delete stale leads, re-verify email addresses, re-crawl company pages.

How it’s built

• Stack: Python, FastAPI, SQLite, SMTP library, web crawling, HTMX UI
• Notable engineering:
  • SMTP verification without sending — connects to the recipient’s MX server, initiates the SMTP conversation, and parses the RCPT TO response. Handles catch-all domains, greylisting, and timeouts.
  • Deep crawl extraction — not just a sitemap scrape but targeted page discovery (linkedin.com/company/ redirect from the site, /about/, /team/, /careers/ patterns) with content extraction from each.
  • Queued retry architecture — leads with unknown SMTP verdicts enter a retry queue with exponential backoff. CRM view shows the verification confidence.
  • SCRUM-based development — built through tracked tasks with explicit completion criteria.
• Architecture: FastAPI → SMTP verifier + web crawler → SQLite → HTMX CRM view. Admin actions trigger re-verification without blocking the UI.

Proof points

• SMTP verification pipeline works — process email addresses through MX handshake without sending mail, with catch-all detection and retry queue.
• Deep crawl extracts real contact context — not just an email but the role, company context, and industry signals.
• SCRUM-tracked build — completed through a task pipeline (task 007: retry unknown verdicts + deep crawl + CRM view) rather than ad-hoc iteration.
• CRM view — structured lead table with verification confidence, enrichment status, and admin controls.

What to show

• Demo: Deploy to a public URL (repo is public). The CRM view with a populated lead table and the SMTP verification results is the hero.
• Visuals needed: the CRM view showing enriched leads with verification scores; the deep-crawl results panel per company; the retry queue showing “unknown → verified” progression.

Cross-links

• The data-enrichment pipeline architecture connects to WEDDING-PRICING-COMPARE (enrich raw vendors → structured DB) and theknot-scraper (enrich raw marketplace data).
• Shares the SMTP protocol dance DNA with SALES-BOT (both leverage low-level internet protocols for business advantage).

Case-study angle

The most “scrappy startup ops” project in the portfolio: Alex built his own ZoomInfo replacement because the real tool costs money and the data is public anyway. SMTP verification without sending mail is a genuinely underused technique, and the deep-crawl pattern shows he thinks beyond “scrape a single page.” It’s a lead-sourcing engine that costs zero monthly recurring and returns verified data.