A simple question, a difficult answer
Choosing a destination for an international conference involves crossing dozens of criteria: hotel capacity, air accessibility, seasonality, costs, sustainability, regulatory context. This information exists but is dispersed, heterogeneous, and often outdated.
DestinAItor, developed by dFakto with PCMA, answers this question in natural language — relying on consolidated, historized, traceable data rather than approximate generative search.
Why a Data Vault foundation under an AI platform
An assistant that recommends a destination must be able to justify its answer: which source, which date, which definition. That's exactly what a Data Vault model provides natively — every record carries its origin and timestamp.
The foundation built with beVault aggregates public and partner sources, applies quality controls upstream of restitution, and exposes a stable semantic layer to the platform's AI components.
- Consolidated and historicized destination data, with source traceability.
- Quality checks applied pre-exposure, not post-reporting.
- A semantic layer shared by the interface and the AI agents.
- Adding new sources without redesigning the existing model.
What the platform brings to its users
More than 12,000 event industry professionals now access a consistent, comparative database, where the work used to rely on spreadsheets and calls to tourist offices.
The gain isn't just time: it's defensibility. A destination recommendation can be backed up in front of a committee with sourced, dated figures.
What this project demonstrates
DestinAItor is a market-facing product, not an internal dashboard. It shows what an industrialized data foundation makes possible: building end-user-facing AI applications without reinventing the data layer for every use case.
The same architecture applies to any sector where answer reliability drives adoption — healthcare, finance, the public sector, industry.