Curation transparency
The shelf report
Most directories tell you how many apps they list. This page tells you how many we turned away. Bloom’s ranking is never for sale, so the only number that proves the curation is real is the one nobody else publishes: the rejection rate. Everything below is generated live from the ingest pipeline.
Last 30 days of ingest
3,647 reviewed → 878 kept
A 76% rejection rate, applied identically to every source. No listing fee exists that can change it.
Shelf composition — 271 live apps
| Creative | 78 |
| Daily Life | 73 |
| Habits | 37 |
| Wellness | 28 |
| Journal | 21 |
| Learning | 16 |
| Agents | 13 |
| Money Mindset | 4 |
| Companion | 1 |
Where supply comes from
| Product Hunt | 138 |
| App Store | 86 |
| Seed | 28 |
| Hacker News | 15 |
| YC F26 | 3 |
| YC S26 | 1 |
141 apps added in the last 30 days. Stale, unloved listings are pruned automatically after 90 days — shelf size is an output, never a goal.
How the gate works
Every candidate passes a two-layer filter: global reject patterns first (B2B and dev tooling, enterprise agents, bank-integrated finance, crypto speculation, NSFW companion bait, content farms, school-procurement software), then word-boundary category matching — no keyword fit, no listing. Rejected supply is counted, not resented: it tells us which categories to open next. See what made it through on the new-apps shelf or the full directory.