AI Transparency

1. AI providers

The current Visibility product integrates OpenAI, Anthropic, Perplexity and Google Gemini APIs. Provider/model configuration may evolve, and historical evidence retains the provider/model identity associated with the recorded observation.

2. How AI is used

• Run tracked prompts and collect provider responses and available citation/source evidence.
• Analyse mentions, recommendations, citations, sentiment, competitors and relative position.
• Generate or prioritise recommendations and readiness/action guidance.
• Support prompt intelligence, competitive intelligence and related analysis where enabled.

3. Data sent to AI providers

A provider request may contain the tracked prompt and configured business context required to perform the measurement. Customer account credentials and payment-card data are not intentionally included in provider prompts. Customers should not submit sensitive personal data in prompts unless a feature expressly requires it and suitable terms and safeguards are in place.

4. Provider training and retention

Treatment of API inputs and outputs depends on the applicable provider terms, product tier and account/configuration in effect for that provider. MonitorMyGEO therefore does not make a blanket statement that every AI provider never retains or trains on every request. Enterprise customers requiring provider-specific contractual assurances should verify the applicable provider terms through the DPA/procurement process.

5. Stored evidence

MonitorMyGEO stores provider/model identity, response evidence, citations where available, timestamps, usage/cost metadata and derived interpretations needed to reproduce and audit historical measurements while the relevant customer workspace remains active. Valid workspace erasure removes customer-owned evidence subject to documented legal exceptions.

6. Limitations

• AI outputs are probabilistic and can vary by model version, time, locale, personalization, retrieval/search availability and provider configuration.
• API observations may not reproduce a consumer chat interface exactly.
• Automated extraction, classification and recommendations may be incomplete or incorrect and should be reviewed against stored evidence.
• Scores and recommendations do not guarantee future ranking, citations, traffic, revenue or commercial outcomes.

7. Automated decisions

MonitorMyGEO uses automated analysis to provide business measurement and decision-support. The product is not designed to make legally significant or similarly high-impact decisions about individuals without human review.

8. Human review and corrections

Customers can inspect underlying evidence where the product exposes it, and material errors should be reported through support. Methodology changes are versioned so historical evidence is not silently rewritten merely because analysis logic evolves.