We said MAYBE on October 22, 2026. Not settled — due October 22, 2027.
Read this with the caveat. We tested the engine that produced this verdict against 292 launches whose outcomes we already knew, and could not show it predicted which survived. Some of its data sources were also dead at the time of scoring. The verdict stays up, dated and unedited, because a record you can quietly revise is not a record — but it is worth less than it looked when it was written.
Real money exists in supply-chain security (4 paying incumbents) and there is a genuine uncovered gap around AI model/dataset provenance, but the evidence for AI-specific demand is thin: zero Reddit, no keywords, and HN threads in single-digit points. This is a plausible OSS-led wedge with a hard monetization path — build it only as a cheap, fast, open-source CLI bet, not as a company thesis.
Four well-funded competitors successfully charge money for supply-chain security, which proves the broad market pays — but the AI-specific slice shows only faint, low-engagement signal and no keyword pull, so demand for the actual wedge is unproven.
A concrete wedge exists — CLI-first, airgapped-capable scanning of non-code AI artifacts (weights, datasets, fine-tuning lineage) that none of the incumbents cover — but every incumbent is a funded giant and the eventual buyer is an enterprise security team gated by procurement, which is the structural barrier here.
The axes say 'maybe' — but the honest case for SKIP is that security tooling almost never monetizes bottom-up: developers install the free CLI, star the repo, and never pay, while the people with budget (enterprise AppSec) only buy through procurement cycles a solo founder cannot survive. The HN signals here are 3-9 points, which is noise, not demand, and 'AI supply chain' risk is currently more a conference talking point than a line item anyone is funding.
What is worth more than this page. The register records what became of 14,962 real launches. No engine has to be right for that to be true.