Methodology — How AistackGraph Works

AistackGraph monitors companies and distills their public activity into six signal channels. This page explains how we collect and score signals. The intelligence is evidence-based and auditable — not guesswork presented as fact.

1. The six signal channels

• News — media coverage and official announcements (Google News RSS).
• Hiring — job postings from public ATS feeds (Greenhouse, Ashby, Lever): total roles, AI-related roles (AI Engineer, ML Engineer, RAG, LLM…), hiring surge detection.
• Product — homepage/pricing/product page content-change detection.
• Tech stack — hosting, frameworks, AI APIs and AI-native docs detected from public pages.
• GitHub — public repository activity and star growth.
• Social — public social mentions (e.g. Weibo keyword feeds via RSSHub).

2. Data sources (all public)

• Company websites, pricing and product pages
• Public documentation and API references
• Job postings and public ATS feeds
• Public GitHub repositories
• News feeds and public social media mentions

We do not access private accounts, private repositories, or non-public information.

3. Signal scoring

• Every signal is stored with a severity (info / notable / major) assigned by an LLM with rule-based fallback: funding, executive changes, launches and pivots are major; hiring surges and pricing changes are notable; routine news is info.
• Duplicate signals are suppressed (same company, type and title within 24h).
• Pages protected by bot defense may yield fewer signals — we report less, never fake more.

4. AI summaries

Signals are summarized by an LLM (a third-party AI provider). Summaries are generated from public signal titles and context only; they are machine-generated and may contain errors. They are provided as orientation, not as professional, investment or legal advice.

5. Freshness

• News is collected daily; hiring, product and tech signals on a recurring schedule.
• Weekly digests aggregate the last 7 days of signals.

6. Corrections

See something wrong? Email [email protected] with the company and the signal you believe is incorrect, and we will re-check and correct it.