Moats
1034 apps · 13 moatsCode stopped being a moat. Every app here is tagged with what's left holding it up — 1 to 3 tags, strongest first.
556of 1034 apps (54%) lean on execution polish — the moat AI erodes
73(7%) have nothing else: polish is the entire defence
The walls still standing
Ranked by catalog frequency. Open one to see which products still depend on it.
- 01execution polishPolish, reliability, sync quality, import fidelity — execution, not structure.
- 02infrastructure scaleInfrastructure one person can't match: global hosting, deliverability, uptime, media pipelines.
- 03integrationsConnector breadth, and the endless upkeep that keeps every connector working.
- 04collaborationIt only pays off once the whole team is in it: shared editing, presence, permissions.
- 05proprietary modelsCustom-trained or frontier models, plus the compute and inference behind them.
- 06content & rightsLicensed content, media rights, curriculum, template and asset libraries.
- 07proprietary dataData you can't rebuild: indexes, crawls, live feeds, archives, maps.
- 08compliance & regulationRegulated ground: licensing, payroll, tax, KYC, HIPAA, real legal exposure.
- 09network effectsIt's better because other people are already on it: graphs, communities, audiences.
- 10brand & trustPeople pay because it's this vendor, and nobody got fired for that.
- 11switching costsYour own accumulated history, config and habits make leaving painful.
- 12marketplace liquidityTwo sides that need each other, and both of them showed up.
- 13hardwarePhysical devices, or data only the vendor's hardware produces.
Tags are part of the dataset, so they get argued about in public ·the definitions live in the repo ↗