​
/v1/public/scrape/site

Use site when you want raw rendered website signals with minimal interpretation. Typical output includes:
  • rendered HTML or extracted text
  • page title and metadata
  • JSON-LD / schema.org data
  • favicon, social links, and basic presentation signals
  • optional subpage expansion
This is the right fit when your own system will do the downstream analysis.

​
/v1/public/scrape/brand-design

Use brand when you want a company’s visual brand identity — the design tokens you’d need to render on-brand emails, cards, or previews — without the full company profile. It renders the homepage (plus about/contact), runs an LLM brand analysis and a vision-based logo pick, then deterministically builds an email-ready palette from the AI’s brand color + the rendered computed colors + screenshot pixel-sampling. Output includes:
  • company name, business type / subtype, and visual tone
  • a full brand palette — primary, secondary, accent, background, surface, text_primary, text_secondary, header_background, footer_background, footer_text, and theme_mode (light/dark) — plus the canonical brand_primary color and a palette_confidence
  • typography (heading/body font families, sanitized to concrete brand fonts)
  • the selected logo in two forms: logo_url (the chosen remote asset) and logo_crop_png (a base64 PNG of the cropped logo region, background baked in), plus favicon_url and og_image
  • social links
  • footer / legal / contact details (privacy, terms, support email, address)
It sits between site and company: heavier than site (it adds the LLM + vision passes and the deterministic palette build), but narrower than company — it does not extract firmographics, pricing, people, or taxonomy. Reach for it when you care about how a brand looks and presents itself.
The palette and theme_mode are governed by the AI’s brand_primary + theme_mode and the rendered computed colors, then rebuilt as a clean neutral canvas — they are not a pure sampling of the site’s own background/surface. Two behaviors differ from a full email-platform pipeline: there is no third-party design-token gap-fill, and logo/header contrast reconciliation (which needs the downloaded logo bytes) is left to the consumer.

​
/v1/public/scrape/company

Use company when you want a higher-level company or brand profile. This path is the heavier interpreted workflow and is intended to combine:
  • homepage rendering
  • key subpage discovery
  • structured company data extraction
  • pricing, people, firmographic, and taxonomy-style enrichment when available
This is the better fit when you want one standardized company object instead of raw page signals. Orchestrators that have a recent pricing-only result may pass its pricing object as options.reusePricing. The company workflow validates the snapshot, keeps it available to summary/taxonomy stages, and skips pricing-page fetch, render, text-model, and vision work. Missing, malformed, or empty snapshots are treated as a cache miss and pricing runs normally. These endpoints complement site and company:
  • /v1/public/scrape/mentions
  • /v1/public/scrape/seo
  • /v1/public/scrape/infra
  • /v1/public/scrape/ads
Use them when you need domain-level visibility around public mentions, SEO posture, DNS/technographics, or ad archive presence.