Last updated June 15, 2026
Meridian is an AI-assisted product, so parts of the experience are powered by AI. AI outputs are checks and suggestions, not decisions: you decide what to do, and so does the consulate. Personal data (passport numbers, dates of birth, financial identifiers) is removed before any content reaches an AI provider.
These are the AI providers we use and exactly what each one does.
| Provider | What it does | Data it sees |
|---|---|---|
| Anthropic (Claude) | Visa guidance, readiness assessment, and drafting application letters | Anonymised queries (personal data removed) |
| OpenAI | Reading uploaded documents and structuring their contents | Document text (personal data removed) |
| Voyage AI | Semantic search across our visa-rules knowledge base | No personal data |
| Jina AI | Extracting visa-requirement content from official web sources | No personal data |
Your data is used to help you, not to train models. We do not train AI models on your conversations, documents, or vault, and our providers are contractually bound not to train on data we send them.
Whether your prior conversations and vault personalise future AI assistance is controlled by the AI personalization toggle on the Cookies page. Core AI features (chat, readiness checks) work whether or not personalization is on.
See also our Privacy policy and Cookies and data page.
