# Depureco USA AI/LLM Website Package — v7 Update Summary

Generated: 2026-06-10T20:05:00Z

## What v7 Adds

v7 is the technical/semantic quality-control layer. Earlier versions built the public database maps. v7 adds the missing pieces needed to deploy, monitor, normalize terms, and evaluate answer quality.

### New Public Files

- `/ai/db/crawler-access-policy-v7.json`
- `/ai/db/structured-data-implementation-map-v7.json`
- `/ai/db/entity-alias-normalization-map-v7.json`
- `/ai/db/spec-measurement-intake-map-v7.json`
- `/ai/db/ai-answer-evaluation-rubric-v7.json`
- `/ai/db/refresh-priority-scoring-map-v7.json`
- `/ai/db/ai-file-observability-map-v7.json`

### New Internal-Only File

- `/internal-review/v7-server-deployment-test-plan.json`

## Main v7 Improvements

1. Separates OpenAI search visibility from model-training crawl policy.
2. Adds explicit noindex/follow handling for `/ai/*.md` and `/ai/db/*.json`.
3. Adds a structured-data implementation map for product, category, application, guide, and AI-support file types.
4. Adds canonical terminology rules for ATEX → ORD LOC/explosion-proof, Longopac as collection not filtration, dust collector vs industrial vacuum, sump vs dry metal chips, and other buyer-language aliases.
5. Adds spec and measurement intake routing for RFQ/product-selection answers.
6. Adds an answer-engine test rubric so ChatGPT/Perplexity/Copilot/Google AI-style answers can be scored after deployment.
7. Adds refresh/build scoring logic to decide whether to update an existing page, build a new page, or add FAQ/internal links.
8. Adds observability fields to monitor whether AI files help or hurt organic traffic and lead generation.

## Upload Rule

Upload public files only:

```text
/llms.txt
/llms-full.txt
/ai-sitemap.xml
/ai/*.md
/ai/db/*.json
```

Do not upload `/internal-review/*`.

## Header Rule

Use `X-Robots-Tag: noindex, follow` for `/ai/*.md` and `/ai/db/*.json`. Do not block `/ai/` with robots.txt if crawlers need to read the files.