DIGITARISE
esc

8 of 79 indexed · ↑↓ to move · ↵ to open

AI platforms website statistics

Every number on this sheet was measured, not surveyed: one HTTPS request per homepage across 33 AI model providers and AI infrastructure sites (35 sampled; 2 excluded, of which 2 refused the request outright), no JavaScript executed.

Rev B — 2026-08-03n = 33dataset (JSON)

The numbers

measured 2026-08-03
120 msmedian server response

Half of the 33 measured homepages return their first byte within this time; the slowest takes 426 ms.

404 KBmedian HTML delivered

Size of the delivered document before any subresource loads. 33 of 33 serve it compressed.

82%ship a known tracker

27 of 33 homepages reference a recognised analytics or advertising tracker in the delivered HTML.

10median third-party domains

The median homepage references this many distinct external domains before any interaction.

39%surface a consent mechanism

Only 13 of 33 expose a detectable consent platform — yet 16 set a cookie before any choice is possible.

85%expose a main landmark

28 of 33 provide a main landmark for assistive technology; 3 ship no first-level heading at all.

Explore every row

live explorer
Median trackers1
≥1 tracker27/33
Any consent signal13/33
  1. 01Copy.ai49/1003rd-party18trackers4pre-cookies1
  2. 02Weaviate57/1003rd-party18trackers5pre-cookies0
  3. 03LangChain60/1003rd-party18trackers1pre-cookies1
  4. 04LlamaIndex61/1003rd-party16trackers2pre-cookies0
  5. 05Jasper62/1003rd-party14trackers2pre-cookies1
  6. 06Replicate63/1003rd-party16trackers0pre-cookies3
  7. 07Glean64/1003rd-party23trackers1pre-cookies1
  8. 08Synthesia67/1003rd-party10trackers2pre-cookies1
  9. 09Baseten68/1003rd-party14trackers1pre-cookies0
  10. 10Stability AI69/1003rd-party19trackers1pre-cookies1
  11. 11Together AI69/1003rd-party13trackers3pre-cookies1
  12. 12Fireworks AI69/1003rd-party13trackers1pre-cookies0
  13. 13ElevenLabs69/1003rd-party13trackers2pre-cookies3
  14. 14Cursor70/1003rd-party6trackers1pre-cookies4
  15. 15Vellum71/1003rd-party6trackers2pre-cookies2
  16. 16Mistral AI72/1003rd-party9trackers1pre-cookies1
  17. 17Harvey72/1003rd-party11trackers1pre-cookies0
  18. 18DataRobot73/1003rd-party7trackers2pre-cookies0
  19. 19Deepgram73/1003rd-party10trackers1pre-cookies0
  20. 20Anthropic75/1003rd-party8trackers0pre-cookies2
  21. 21Runway76/1003rd-party11trackers1pre-cookies3
  22. 22Weights & Biases76/1003rd-party14trackers1pre-cookies1
  23. 23Databricks76/1003rd-party14trackers1pre-cookies1
  24. 24Chroma77/1003rd-party7trackers1pre-cookies0
  25. 25Hugging Face78/1003rd-party9trackers0pre-cookies0
  26. 26Groq79/1003rd-party10trackers2pre-cookies0
  27. 27H2O.ai79/1003rd-party10trackers2pre-cookies0
  28. 28Scale AI82/1003rd-party3trackers1pre-cookies0
  29. 29Modal82/1003rd-party6trackers0pre-cookies0
  30. 30Pinecone83/1003rd-party10trackers1pre-cookies0
  31. 31Humanloop85/1003rd-party4trackers0pre-cookies0
  32. 32Cohere88/1003rd-party6trackers1pre-cookies0
  33. 33AssemblyAI97/1003rd-party2trackers0pre-cookies0

33 of 35 AI model providers and AI infrastructure homepages · measured 2026-08-03 · score = transparent composite of the raw signals in each row

The statistics above are recomputable from this table — switch dimensions to re-rank the sector.

Against the other sectors

identical method

Share of homepages referencing a recognised analytics or advertising tracker in the delivered document.

B2B SaaSn = 3385%
AI platformsn = 3382%
Developer toolsn = 3376%
Fintechn = 2767%
Healthcaren = 2952%
DTC ecommercen = 2492%
Retail chainsn = 1872%

Healthcare holds the best posture here — 52% against 85% and 82% and 76% and 67% and 92% and 72%.

Lower is better · all sectors measured with the same script and method · 2026-08-03

Every measured sector, one metric at a time. The ordering changes with the metric — which is the point.

Every exclusion, named

2 of 35
OpenAI — HTTP 403 (automated request blocked)Perplexity — HTTP 403 (automated request blocked)

Cite this

CC BY 4.0

Free to reuse with attribution. Quote the revision, not the date you read it — Rev B is a specific set of measurements, and re-measurement publishes a new one rather than editing this.

Reference
Digitarise. “AI platforms website statistics.” Rev B, measured 2026-08-03. n = 33 of 35 sampled. https://www.digitarise.com/data/ai-website-statistics
BibTeX
@dataset{digitarise_ai_platforms_2026,
  title     = {AI platforms Website Observatory},
  author    = {{Digitarise}},
  year      = {2026},
  version   = {Rev B},
  note      = {n = 33 of 35 sampled; measured 2026-08-03; one HTTPS GET per homepage, no JavaScript executed},
  url       = {https://www.digitarise.com/data/ai-website-observatory.json},
  license   = {CC BY 4.0}
}

The raw measurements are a downloadable JSON file — every row, every excluded site, and the reason each was excluded.

One request per homepage. No JavaScript executed, nothing simulated — only what any client can observe.

Read the methodology