Find very early startups and founders who could give an investor their next meeting a round earlier: search the public web broadly for idea-stage, stealth, pre-product, pre-seed and unusually early seed companies, prioritizing the United States and Canada; require public evidence connecting founders to a real product, research effort, team or company; rank up to 50 strong leads without padding; and exclude every company on the supplied Scout list.
Scout list not received. The request contained only a placeholder where the Scout company list should have been pasted. These leads have not been compared with the Scout list and must not be treated as net-new or Scout-excluded. Provide the full list for a deduped update. This output is discovery, not diligence: each lead row rests on a single public source, some of them founder- or company-reported — verify founder employment, incorporation, funding, customer claims and profile ownership before outreach.
Broad public-web discovery returned 18 ranked rows, not 50 — the ranking was capped at strong leads rather than padded. 17 are active prospects (11 North American, 4 international, 3 with no stated location among them) and one, Invariant Labs of Zurich, was acquired by Snyk and is kept only as a labeled control. Fit scores run 0.749–0.768; earliest public signals span 2024 to mid-2026.
Each card separates what the source states (facts as reported, not independently verified) from why it may fit venture pre-seed (analysis). Cards follow discovery rank. Null profile links and funding stay marked as not stated.
| # | Company | Location | Founders | Stage / funding | Signal | Conf. | Fit | Source |
|---|
Method: 18 rows from one broad public-web discovery pass ranked by semantic fit (0.749–0.768), signals dated 2024 to Sep 2026, priority 2025–2026 and US/Canada. Fit score measures semantic match to the brief, not quality or diligence. Each row is single-sourced; 16 distinct source hosts, some founder posts, company pages, aggregators or syndications. Scout-list exclusion was not performed — the list was never supplied. Long evidence and analysis text is truncated on cards; the linked sources hold the full statements. No endorsement or independent verification is implied.