Two-month public-web scan surfaces 1,125 people engaging with decision models — 18.75× the prior shortlist, not the asked-for 100×
Asked (summary):Find the engineers, technologists and researchers inside large AI-active enterprises, startups and scale-ups worldwide — Europe, US, India, Israel in particular — who publicly expressed interest in decision models (TypeSafe Jev, Perplexity Decider and the like) on LinkedIn, X, blogs, GitHub, Hugging Face or elsewhere in the last month.
Asked (follow-up):“I wanted 100x this volume dude, 100x more people, 100x more companies. READ EVERYTHING anyone is saying on this topic in the past month or two, this is NOT ENOUGH!”
The expanded scan — five families of searches, over 11,000 gross search-result records before deduplication — yields 1,125 distinct people or public handles with attributable decision-model signals dated September 3 to October 30, 2026, drawn from 671 URLs on 360 hosts. 477 rows state an affiliation and 328 distinct organizations are named. That is 18.75× the prior 60-person shortlist and roughly 12× its 27 organizations, but honestly short of a literal 100× (6,000 people): non-indexed likes, private LinkedIn/X activity, missing employer data and the category's recent emergence cap what the public web can defensibly attribute. This is a high-recall lead universe, not “everything anyone said”.
Each dot is one person or handle, placed by signal date and semantic signal strength; colour is the model family they engaged with. Click a legend entry to toggle a family; filters and search re-draw both the chart and the table. Hover a dot for the person, affiliation and the exact public signal.
TypeSafe Jev / System One dominates: 649 of 1,125 signals (58%) concern Jev directly, against 71 for Perplexity Decider and 23 each for Microsoft Decision-1 and Cloudflare Clef — e.g. the Bonn/Lamarr/Fraunhofer team's independent zero-shot evaluation of jev-1.13.0 over 37 datasets and 346,009 requests. local-ai-zone.github.io
Hands-on evidence is substantial: 150 GitHub signals, 92 Hugging Face and 87 arXiv/pre-print rows are builders, evaluators and publishers rather than commenters — like the ejs-5/jev-benchmark evaluating Jev on 868 real decisions. bugtraqsolutions.com
Enterprise names recur at the top of the affiliated slice: OpenAI and Microsoft (8 rows each), Stanford (7), TypeSafe, Perplexity, Vercel and UIUC (6 each), Cloudflare and Hugging Face (5 each) among 328 named organizations. huggingface.co
Stated geography is sparse, not absent interest: only 3 rows text-match India, 2 Israel and 2 Canada in organization/location fields — these count explicit statements, never inferred nationality or actual regional coverage. alikoker.com.tr
All 1,125 people and handles
#
Person / handle
Organization · role
Model / topic
Public signal
Date
Platform
Strength
Source
Public-web scan for decision-model interest signals, 2026-09-03 to 2026-10-30. 1,125 deduplicated rows (one per lowercased name or handle) from 671 URLs on 360 hosts; largest single URL contributes 15 rows (1.33%), largest host 169 (15.02%). Signal strength and topic relevance are semantic scores (0–1), not prestige. Blanks preserved: employers, roles, locations and identities behind handles are never inferred; secondary pages may quote an original action; every lead needs human verification before outreach. Conservative handle-based deduplication may keep one human twice or merge common one-word names. Table paginates for space; the CSV download carries every filtered row.