
The GTM Intelligence Platform transforms raw public web signals into actionable B2B sales intelligence — detecting when companies adopt new enterprise software long before any public announcement. Stage 1 — Signal Collection runs four concurrent collectors: a real-time Certificate Transparency log stream, a multi-source DNS subdomain harvester from six free APIs, a Wayback Machine CDX crawler detecting new integration pages, and a GitHub activity monitor flagging commit spikes and customer mentions. Stage 2 — Bright Data Integration is the backbone of the web data layer. The main pipeline uses the Bright Data REST API (SERP zone) for budget-guarded SERP queries and page rendering. The Bright Data MCP server is used directly via SSE, calling search_engine and scrape_as_markdown tools for live web intelligence. Stage 3 — Parsing & Normalization routes signals through specialized parsers and a VendorFingerprinter with 36+ pre-computed patterns (Salesforce, HubSpot, Stripe, Okta, Snowflake, Datadog) to assign vendor hints and confidence scores. Stage 4 — Correlation Engine groups signals into DealCluster objects using a pandas 30-day rolling window and a NetworkX bipartite graph. A multi-factor scorer assigns each cluster a HIGH, MEDIUM, or LOW confidence tier. Stage 5 — AI Enrichment feeds each cluster into GPT-4o-mini, returning the suspected vendor, deal close date, optimal outreach window, and reasoning. Confidence is blended 60% LLM + 40% Stage 4. Stage 6 — Delivery exposes intelligence through a FastAPI REST API, React 18 dashboard, PostgreSQL storage, email via Resend, and Slack alerts for HIGH-tier deals. Deployed on Render via Docker. Launch Sniper is a companion module detecting unreleased competitor products via WHOIS registrations, USPTO trademark filings, and robots.txt changes — using the Bright Data MCP server for live scraping and search, then generating AI-powered counter-playbooks delivered via email and Slack.
31 May 2026