ai.txt

# TH Analytica Framework: ai.txt

This document summarises the core guidelines for AI Visibility and AI Readiness within the TH Analytica Framework.

### AI Visibility Guidelines

- **Structured Data**: Encode applicable, visible and verified information with standards such as Schema.org and JSON-LD. Structured data can reduce ambiguity but does not guarantee use by an AI product.
- **Semantic Consistency**: Use clear, unambiguous terminology across your site. Align your wording with commonly used concepts in your domain to reduce semantic friction.
- **External References**: Provide natural, relevant citations to authoritative sources. Independent corroboration can support verification, but provider behaviour varies.
- **Balanced Optimization**: Avoid over-optimizing solely for AI systems. Maintain readability for humans and naturalness for search engines to prevent algorithmic downgrading.

### AI Readiness Guidelines

- **Process Clarity**: Document your key business processes and workflows in a way that can be consumed by AI agents. Clear steps and roles reduce friction.
- **Entity Linking**: Map your organisation’s key entities (people, products, services) to consistent identifiers so AI can reference them correctly.
- **Agent Interoperability**: Where appropriate, expose documented APIs or standard formats. Do not imply permission for autonomous transactions.
- **Security & Governance**: Implement appropriate access controls and governance mechanisms to protect sensitive data while enabling AI operation.

For detailed examples and definitions, consult the framework documents and definitions.md in this repository.

### Boundaries

- Crawler access is not permission for model training.
- A public corpus record is not proof of model training.
- Model training is not proof of query-time retrieval.
- Retrieval does not guarantee citation, mention or recommendation.
- Files such as `ai.txt` and `llms.txt` are informational and have provider-dependent support.
- Consequential actions require explicit human approval.

Canonical website: https://th-analytica.com