What kind of data VoyantIQ holds, and how AI actually touches it.
This page describes VoyantIQ's actual current data classification and AI practices in plain language. It is not a substitute for professional legal review. See Privacy for what VoyantIQ collects and what you can see and correct.
How VoyantIQ Classifies Data
Every category below is a real classification this codebase actually enforces, not an aspirational policy.
Personal Information
Your name, email address, and account identity. Used to operate your account and never sold.
Trip and Customer Content
What you tell VoyantIQ about a trip: destination, dates, budget, travelers, preferences, and the messages you send while planning. Used to power recommendations for that trip and, once you confirm a preference, to avoid asking you the same thing again.
Sensitive or Higher Risk Information
Accessibility needs, medical or safety notes, and travel document detail. Held to a stricter standard than the rest of your Trip and Customer Content: restricted fields never appear in a list view or the default agent projection AI reasoning reads from, and this codebase never stores a full document number, a document image, or a payment credential at all.
Decision Data
The recommendations VoyantIQ produced, the decisions you made about them, and the Learning Events that come from a real, well-defined trigger, such as locking a decision or stating a preference. Used to improve the recommendations you personally see going forward.
Aggregated or Deidentified Insights
Plain counts and rates computed over Decision Data, such as how often a category's top recommendation is actually chosen. Today this is scoped to a single workspace's own data only; VoyantIQ does not compute or store any insight that combines data across separate workspaces or customers.
Commercial Insight Products
A future category, not a current one. VoyantIQ does not build, sell, or offer any product based on aggregated or deidentified customer data today. If that ever changes, it happens only after deidentification, a minimum cohort size large enough to protect any individual traveler's privacy, and explicit founder approval before the capability is built, not after.
Where AI Is Actually Involved
AI helps interpret a planning conversation into a structured request and helps phrase an explanation. It never makes the underlying decision by itself: constraint checking, candidate evaluation, ranking, and explainability are VoyantIQ's own deterministic reasoning, not something an AI model is trusted to get right on its own.
Provider Independence
External AI models are reached only through VoyantIQ's own routing layer, never called directly from application code. If a configured provider is unavailable, VoyantIQ falls back to another configured provider, and ultimately to a simple, deterministic response, rather than failing outright.
What Never Reaches Analytics or Telemetry
No prompt, no AI response, and no personal or sensitive content ever reaches analytics or AI operational telemetry, not even indirectly. Operational telemetry about an AI request captures only metadata, such as which provider responded and how long it took, never the content of the request or its response.
Your Data Is Not Used to Train a Shared Model
VoyantIQ does not use your Trip and Customer Content to train a model shared across other customers. If that ever changes, it is a Commercial Insight Product by this page's own definition above, and requires the same founder approval and deidentification standard before it happens.