AI Engines
Which AI engines does GeoScript monitor?
GeoScript runs real API observations against live AI engines using deterministic prompt sets, then preserves raw responses before parsing. This page documents exactly what GeoScript observes from each engine — and what it does not.
Real API observations, production adapters, smoke-tested
ChatGPT
GeoScript monitors how ChatGPT responds to commercial, local, and comparison queries that affect your business — tracking mentions, recommendation rate, and response text across prompt categories.
See coverage →Claude
GeoScript monitors how Claude responds to business and commercial queries, tracking mentions, recommendation rate, and response content across your prompt universe.
See coverage →Perplexity
GeoScript monitors how Perplexity responds to business queries and — uniquely — captures the structured citations Perplexity natively reports, giving you the highest-signal citation intelligence of any monitored engine.
See coverage →Not yet live — no production observations exist
Gemini
Gemini integration is in development. The adapter is written and includes support for grounding metadata — structured citation sources Google returns when search grounding is active. Live production monitoring is not yet available.
See coverage →Google AI Overview
Google AI Overviews appear in Google Search results as AI-generated summaries. GeoScript is developing integration to monitor when businesses appear in these summaries. Live monitoring is not yet available.
See coverage →Microsoft Copilot
Microsoft Copilot integration is in development. GeoScript plans to monitor how Copilot responds to commercial and local business queries. Live monitoring is not yet available.
See coverage →How GeoScript observations work
For every monitored engine, GeoScript runs a deterministic prompt set specific to your business — your industry, services, market, and competitive context. Each prompt is submitted as a real API call to the engine's production endpoint.
The raw HTTP response body is preserved before any parsing. This means the observation record reflects what the engine actually returned, not an interpreted version. GeoScript then extracts business mentions, recommendation signals, and citations from the raw response.
Observations accumulate over time. You can see how each engine's responses to your prompt universe change across runs — not just a point-in-time snapshot.
What GeoScript does not measure
These limitations apply across all engines. Each engine's detail page lists engine-specific limitations.
- ✕Dwell time or user engagement. GeoScript cannot observe what a user does after reading an AI response.
- ✕Internal model ranking signals. GeoScript does not have access to any engine's internal parameters, training data, or ranking logic.
- ✕User-level personalization. GeoScript runs API-level observations. Individual users may receive different, personalized responses.
- ✕Future response guarantees. An observation confirms what an engine said at a point in time — not what it will say in the future.