AnomalyArmor supports two types of tags: auto-classification that detects PII patterns automatically, and custom tags you create to match your organization's needs.
## Tag Types
| Type | What It Does | Use For |
| ------------------- | ------------------------------------ | ------------------------------- |
| Auto-classification | Detects PII patterns in column names | Compliance, PII tracking |
| Custom tags | Labels you define and apply | Organization, ownership, status |
## Common Questions
### Does auto-classification read my actual data to detect PII?
No. Auto-classification runs pattern matching against column names, types, and AI-generated profiles - never row values. So a column named `user_email` is classified as PII-email from its name, not from a data scan. See [auto-classification](/data-classification/auto-classification) for the full detection model.
### How do I create a custom tag for something like 'team ownership'?
Create a tag set in **Settings → Tags** (e.g., `team:growth`, `team:platform`) then apply it to assets manually, via API, or as a rule (tag any asset in a specific schema with a specific team). See [custom tags](/data-classification/custom-tags).
### Can I use tags in alert rules to scope notifications?
Yes. Alert rules support tag-based filtering - route all `pii:email` asset alerts to the security channel, or page only on `tier:critical` tables. This is typically how teams avoid noisy alerting across heterogeneous data.
### Are auto-classification tags guaranteed accurate?
They're high-precision heuristics, not a substitute for your own review. Cases like columns with ambiguous names (`id`, `value`, `data`) are more likely to be missed or mis-tagged. Treat auto-tags as a starting point for compliance review, not the final word.
## Next Steps