BASE
Born from crawled page content, not sitemap alone
BASE
AI language on real pages, after sitemap — never from the map alone.
Journey B: enroll, map URLs, crawl content, then write a Web AI row only when a page clears the quality bar — product claims with substance, not vague AI marketing. Sitemap mapping alone never creates a signal.
- Sitemap / path map never creates signals
- Rows come from page content after crawl + tier check
- Multiplicity is by URL / page kind, not hiring-style depth
- Structural (100) or claim_with_metric (75) — otherwise none
Sample signalBASE
Web AI · /product/agents
Journey B · map never writes · sample
Enroll
Company enters pipeline
Sitemap
URL inventory only
Crawl + tier
Content + quality bar
Write row
WEB_PAGE_AI_ACTIVITY
productchangelogcareersblog_post
BASE
Role
B · crawl
Journey
75 / 100
Tiers
At a glance
How this signal shows up in the store
B
Discovery pipeline: enroll → map → crawl → tier
2
Quality grades that ship: structural and claim_with_metric
URL
One qualifying page can earn its own timeline row
What it is
Evidence that a specific page shows meaningful AI language
A Web AI signal is evidence that a specific company page shows meaningful AI product or ops language — not vague “AI-powered” marketing. Unlike Hiring (many openings → one consolidated depth-scored row), Web AI’s multiplicity is which pages you watch: different URL kinds can each earn their own row.
Journey B always maps the site first. The first successful sitemap is a baseline URL inventory with no signals. Timeline rows appear only after content crawl, an AI-language prefilter, and a quality tier check on first fetch or a real content change.
- Monitored kinds include product, features, changelog, about, and careers
- Blog and news roots discover children; they are not themselves the signal source
- Same URL + content hash already signaled → skip
Signature story
Discovery pipeline + page kinds
Sitemap never creates signals. Rows come from crawled page content that clears the quality bar — by URL kind.
Enroll → map → crawl → tier → row
Journey B
Enroll
Company enters web pipeline
Sitemap
URL inventory only
Crawl
Content fetch + hash
Prefilter
AI-language soft gate
Tier
structural · claim_with_metric
Write row
WEB_PAGE_AI_ACTIVITY
Page kinds
Monitored
product · features · changelog · about · careers
Re-crawled on a cadence
Discovery roots
blog · news · updates
Find children — not the signal source
One-shot children
blog_post · similar
Usually crawled once after discovery
Quality bar
structural 100 · claim_with_metric 75
none → no timeline row
How it forms
Discovery pipeline, step by step
Web AI is a discovery journey: mapping comes first; timeline rows come only later from content that clears the bar.
- 1
Enroll the company
Companies with hiring or funding activity are typically enrolled into the web pipeline and may get crawl priority bumps when other signals fire.
- 2
Sitemap / path map
Build or refresh monitored and discovered URLs. No signal is written here — the first successful run is baseline inventory only.
- 3
Content crawl + prefilter
Crawl budgeted pages (prefer high priority; skip discovery roots). Soft AI-language prefilter stops pages that do not look AI-related. Unchanged content hash updates timestamps only.
- 4
Tier and write
On first crawl or content change, an LLM grades the page. Structural or claim_with_metric writes one WEB_PAGE_AI_ACTIVITY row for that URL. Tier none writes nothing.
Confidence
Two-tier quality bar
Unlike Hiring’s depth ladder, Web AI uses a shared two-grade bar across pages — keep or discard.
BandConfidenceNote
- structural100Dedicated product / case-study style evidence
- claim_with_metric75Specific AI claim that includes a metric
- none—No row written
Stock vs flow
Web AI vs “mentions AI on the homepage” lists
Keyword scrapes treat every AI phrase as intent. Web AI waits for mapped URLs, real content changes, and a quality tier before writing a row.
Layer
Traditional
VeilStrat
Discovery
Before any row
Scrape homepage once
No URL inventory discipline
Sitemap baseline first
Map never creates signals alone
Quality
What counts
Any AI buzzword
Marketing fluff looks like product
Structural or metric-backed claim
Two-tier bar; none discards
Unit
Granularity
One company-level AI flag
Loses which surface fired
Per URL / page kind
Product vs changelog vs post
Change
When it refreshes
Quarterly re-scrape
Misses mid-cycle launches
First crawl or content hash change
Same hash → skip rewrite
Discovery
Before any row
Traditional
Scrape homepage once
No URL inventory discipline
VeilStrat
Sitemap baseline first
Map never creates signals alone
Quality
What counts
Traditional
Any AI buzzword
Marketing fluff looks like product
VeilStrat
Structural or metric-backed claim
Two-tier bar; none discards
Unit
Granularity
Traditional
One company-level AI flag
Loses which surface fired
VeilStrat
Per URL / page kind
Product vs changelog vs post
Change
When it refreshes
Traditional
Quarterly re-scrape
Misses mid-cycle launches
VeilStrat
First crawl or content hash change
Same hash → skip rewrite
Example snapshots
What a Web AI row looks like
Illustrative page fires — real rows attach URL, page kind, content hash, and tier.
Product surface
Web AI · /product/agents
Structural evidence on a dedicated product page after crawl and tier check.
Kind: product · Tier: structural · Confidence 100
Metric claim
Web AI · /changelog
Specific AI claim with a metric clears claim_with_metric on a monitored page.
Kind: changelog · Tier: claim_with_metric · Confidence 75
No ship
Sitemap baseline only
URL inventory refreshed — still no timeline row until content qualifies.
Map step · No WEB_PAGE_AI_ACTIVITY written
GTM moments
Public product language before the shortlist
Use Web AI when published site language is the earliest buying signal you can trust.
Timing
Catch launches before RFPs
Product and changelog pages often fire months before a vendor shortlist forms.
Context
Know which surface spoke
Page kind and URL in the snapshot give reps a concrete citation for outreach.
Enrich
Refresh slices when pages fire
Web AI can soft-refresh notebooks and trigger a targeted hiring pull for the company.
Filter
Ignore vague AI marketing
The two-tier bar keeps fluff off the timeline so lists stay usable.
Journey B
Sitemap never creates signals
The first successful map is a baseline of URLs. Signals come from page content — first fetch or a real content change — after an AI-language prefilter and a quality tier check. Different URLs can each earn their own Web AI row for the same company. Recency decay still applies later in rankings; that is aging, not the journey definition.
Related signals
How this connects on the timeline
Public product language before the shortlist — explore the signals that feed or follow this type.
FAQ
Web AI activity, answered
How Journey B, page kinds, and quality tiers work.
It is a BASE timeline row showing that a specific company page cleared an AI-language prefilter and a quality tier (structural or claim_with_metric) after content crawl — not after sitemap mapping alone.
Read what companiespublish about AI.
Web AI signals attach to real URLs — after discovery and crawl — so you act on product language, not sitemap noise.
© 2026 VeilStrat. All rights reserved.All systems operational