> For the complete documentation index, see [llms.txt](https://zeni.gitbook.io/zeni/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://zeni.gitbook.io/zeni/about-zeni/background.md).

# Background

AI systems are only as good as the data they learn from. The most valuable inputs are not static datasets, but live signals — what happened, when it happened, and whether the event is trustworthy. Today these signals are spread across online platforms and on-chain activity, fragmented across sources, and increasingly polluted by spam, bots, and synthetic behavior.

**Teams building AI, analytics, and automation face the same recurring bottlenecks:**

● Inconsistent schemas across sources

● Unclear provenance for individual signals

● Low reproducibility of training and evaluation pipelines

● Weak integrity guarantees for downstream use

Users and partners also need clear consent boundaries for online sources, and auditability for high-stakes use cases.