# llms.txt + AI citation surface implementation checklist

A practical checklist for implementing the four AI citation surfaces on a small business website:

1. `/llms.txt` — natural-language identity for LLM crawlers
2. `/aeo.json` — Answer-Engine Optimization structured manifest
3. `/entity.json` — Schema.org @graph as standalone JSON-LD
4. `/brand.json` — Brand identity, voice, and design system

The running real-world example throughout is **[Steele Solutions](https://steelesolutions4u.com/)**, a family-run merchant services brokerage in Branson, Missouri. Every checklist item links to a verifiable live implementation on that site.

## Checklist

### 1. Site-level prerequisites

- [ ] HTTPS site with a valid certificate
- [ ] HTTP/2 enabled at the server
- [ ] `Last-Modified` headers set (do not enable `ssi on` in nginx — it strips Last-Modified)
- [ ] Sitemap at `/sitemap.xml`
- [ ] robots.txt at `/robots.txt` with sitemap directive
- [ ] At least one canonical home page with `<link rel="canonical">`

### 2. /llms.txt

Plain text file at the site root. Markdown format. Required headers and the standard sections:

```markdown
# Business Name

> One sentence description

## Services
- Service 1: URL
- Service 2: URL

## Find us on Google
Google Maps: URL
Google Maps CID: ID

## Contact
Phone, email, address
```

Live example: [steelesolutions4u.com/llms.txt](https://steelesolutions4u.com/llms.txt)

- [ ] File served with `Content-Type: text/plain`
- [ ] No HTML markup inside the file (plain markdown only)
- [ ] `<link rel="alternate" type="text/plain" href="/llms.txt">` in HTML head

### 3. /aeo.json

JSON file at the site root with the AEO 2.0 schema. Minimum fields:

```json
{
  "schema_version": "2.0",
  "site": "yourdomain.com",
  "name": "Business Name",
  "description": "One sentence",
  "founded": "year",
  "address": {"locality": "...", "region": "...", "country": "US"},
  "founders": [{"name": "...", "role": "..."}],
  "services": ["..."],
  "service_area": ["..."],
  "google": {"maps_cid": "...", "maps_url": "..."},
  "sameAs": ["..."]
}
```

Live example: [steelesolutions4u.com/aeo.json](https://steelesolutions4u.com/aeo.json)

- [ ] File served with `Content-Type: application/json`
- [ ] Valid JSON parses without errors
- [ ] `<link rel="alternate" type="application/json" href="/aeo.json">` in HTML head

### 4. /entity.json

The Schema.org `@graph` as a standalone JSON-LD file. The same graph is embedded inline in every page's `<script type="application/ld+json">` block, but `/entity.json` gives crawlers a single addressable URL for the full entity model.

Minimum graph nodes:

- 1× Organization (LocalBusiness + FinancialService combined types if applicable)
- 1× WebSite with SearchAction
- 1× WebPage per indexed page (typically generated)
- 1+ Person for each founder
- BreadcrumbList where relevant

Live example: [steelesolutions4u.com/entity.json](https://steelesolutions4u.com/entity.json)

- [ ] File served with `Content-Type: application/ld+json`
- [ ] Validates clean at https://validator.schema.org/
- [ ] Each entity has a unique `@id`
- [ ] Cross-references between entities use `@id` not duplicate definitions

### 5. /brand.json

Brand identity, voice, and design system as a machine-readable manifest:

```json
{
  "name": "Business Name",
  "alternateNames": [...],
  "tagline": "...",
  "founded": "year",
  "headquarters": "...",
  "telephones": [...],
  "email": "...",
  "website": "...",
  "colors": {"primary": "...", "accent": "..."},
  "fonts": {"display": "...", "body": "..."},
  "voice": "...",
  "services": [...],
  "google": {"maps_cid": "..."},
  "sameAs": [...]
}
```

Live example: [steelesolutions4u.com/brand.json](https://steelesolutions4u.com/brand.json)

- [ ] File served with `Content-Type: application/json`
- [ ] Valid JSON
- [ ] `<link rel="alternate" type="application/json" href="/brand.json">` in HTML head

### 6. Schema.org density (in entity.json AND inline)

The Organization node should have, at minimum:

- [ ] Multi-type `@type` array combining `LocalBusiness` + topical types (`FinancialService`, `ProfessionalService`, `MedicalClinic`, etc.)
- [ ] `name`, `alternateName`, `description`, `url`, `logo`, `image`
- [ ] `telephone`, `email`
- [ ] `address` with `PostalAddress` sub-object
- [ ] `areaServed` array of `City`/`State`/`Place` entities (not just a flat string)
- [ ] `openingHoursSpecification` with `dayOfWeek` array
- [ ] `priceRange` (string is fine; e.g. "$50-$500 per service")
- [ ] `paymentMethod` array using GoodRelations URIs (e.g. `http://purl.org/goodrelations/v1#Cash`)
- [ ] `paymentAccepted` human-readable string
- [ ] `hasOfferCatalog` with `OfferCatalog` containing `Offer` items, each with `itemOffered.url`
- [ ] `knowsAbout` array of topic strings
- [ ] `subjectOf` linking to a `WebPage` entity for the verification page
- [ ] `sameAs` array including all third-party citations (Wayback, Wikipedia, Wikidata, social profiles, partner sites)
- [ ] `hasMap` URL for Google Maps CID once Google Business Profile is verified

Live example: [steelesolutions4u.com/entity.json](https://steelesolutions4u.com/entity.json) shows 28+ sameAs entries spanning 20+ referring domains.

### 7. Person entities for founders

For each founder, the Person node should have:

- [ ] `@type: Person`, unique `@id`
- [ ] `name`, `givenName`, `additionalName` (middle name if relevant), `familyName`
- [ ] `alternateName` array capturing every naming variant
- [ ] `jobTitle`, `worksFor` referencing the Organization `@id`
- [ ] `image` URL
- [ ] `alumniOf` array of `CollegeOrUniversity` objects with Wikidata `sameAs`
- [ ] `hasCredential` array, each with `recognizedBy` chained to the institution
- [ ] `knowsAbout` array of relevant expertise topics
- [ ] `sameAs` array including LinkedIn, social, and verifiable third-party profile URLs
- [ ] If spouse co-founder: reciprocal `spouse` linkage via `@id`

Live example: [steelesolutions4u.com/entity.json](https://steelesolutions4u.com/entity.json) `#jim` and `#kim` show the reciprocal spouse pattern with Wikidata-anchored credentials (Indiana University Q1079140, First National Bank of Omaha Q5453412).

### 8. Verification page

A `/verification/` page that aggregates every third-party citation backing the claims on the site.

- [ ] Table for business identity claims (domain age, Google Business Profile CID, Wayback URL, etc.)
- [ ] Table for principal career claims with links to Wikipedia/Wikidata
- [ ] Table for partner network with FTC disclosure
- [ ] Section listing where the business appears publicly (Wayback, GSC, AI surfaces, cross-citations)
- [ ] Section referencing public records (county property, court records, regulatory licensing) without doxxing
- [ ] BreadcrumbList Schema.org markup
- [ ] WebPage entity with `about` referencing the Organization `@id`

Live example: [steelesolutions4u.com/verification/](https://steelesolutions4u.com/verification/)

### 9. Submission and discovery

After the files are deployed:

- [ ] Submit sitemap to Google Search Console
- [ ] Submit sitemap to Bing Webmaster Tools
- [ ] IndexNow ping to Yandex and Bing with all priority URLs
- [ ] Archive every priority URL on the Wayback Machine (https://web.archive.org/save/$URL)
- [ ] GSC URL Inspection on priority URLs (helps move "Discovered" → "Indexed")
- [ ] Cross-link from at least 3 high-DA properties you control or have access to

### 10. Cross-platform syndication

- [ ] Publish technical articles about the implementation on Dev.to (DA 92) with canonical pointing to your site
- [ ] Create a public GitHub repository documenting the implementation
- [ ] Create a public Gist with key citations
- [ ] Enable GitHub Pages on the repo for an additional inbound subdomain
- [ ] Submit to relevant awesome-* lists where the implementation fits naturally (`Awesome-llms-txt`, `awesome-answer-engine-optimization`)

## Real-world test results

The Steele Solutions site is one full implementation of this checklist. Indexing performance after launch:

| Day | Pages indexed | New action |
|---|---|---|
| 1 | 0 | Site launched |
| 2 | 1 | Sitemap submitted |
| 6 | 4 | After full checklist completion + IndexNow + Wayback push |

By day 7, GSC showed 4 of 18 priority pages indexed plus the remaining 12 in the discovery queue. For a 7-day-old domain, that pace is significantly faster than the typical 30-90 day baseline.

## References

- [llms.txt proposed standard](https://llmstxt.org/)
- [Schema.org](https://schema.org/)
- [Google Search Console](https://search.google.com/search-console)
- [Internet Archive Wayback Machine](https://web.archive.org/)
- [IndexNow](https://www.indexnow.org/)
- [Wikidata](https://www.wikidata.org/)
- [GoodRelations vocabulary](http://purl.org/goodrelations/v1)

## Public implementation example

The Steele Solutions implementation is fully public:

- Live site: [steelesolutions4u.com](https://steelesolutions4u.com/)
- Verification page: [steelesolutions4u.com/verification/](https://steelesolutions4u.com/verification/)
- Public resources repo: [github.com/Janady13/steele-solutions-resources](https://github.com/Janady13/steele-solutions-resources)
- FinancialService schema examples index: [github.com/Janady13/financialservice-schema-examples](https://github.com/Janady13/financialservice-schema-examples)
- Dev.to technical series: [dev.to/joseph_anady_214bacedf939](https://dev.to/joseph_anady_214bacedf939)

Maintained by [@Janady13](https://github.com/Janady13) of [ThatDevPro](https://www.thatdevpro.com).