Frequently asked questions

SovraRAG (also written Sovra RAG) is a sovereign retrieval-augmented generation (RAG) platform for enterprise document intelligence. It turns your document archive into an AI analyst that answers questions with citations — using hybrid search, a knowledge graph and agentic reasoning — entirely on Swiss infrastructure, encrypted so that even the operator cannot read your data.

Correct. AI inference and embeddings run on a Swiss AI cloud, compute and object storage run in a Geneva datacenter, and backups stay in the same Swiss bucket. No US-incorporated service ever touches your document content.

Every document is encrypted with AES-256-GCM using a key unique to your organisation before it ever reaches storage. Without that key, not even the platform operator can decrypt your content.

GraphRAG extends RAG with a knowledge graph of entities and relations extracted from your documents, so the system can answer multi-hop questions that span many documents — e.g. “which suppliers are affected by this clause change?”. See our full explainer: What is GraphRAG?

For complex questions, an autonomous but strictly bounded research process investigates each part of your question across the corpus, checks what is still missing and then synthesises one cited answer. Details: What is agentic RAG?

German, French, Italian and English natively. You can ask in one language and retrieve from documents written in another — SovraRAG bridges Switzerland's languages out of the box.

PDF, DOCX, XLSX, PPTX, plain text and scanned images with OCR. Structure, tables and headings are preserved — so answers respect how your documents are organised.

SovraRAG is architected for both: all processing stays in Switzerland (no cross-border disclosure), data is encrypted per tenant before storage, access is controlled per document, retention is configurable, and complete audit and query logs support accountability. See RAG under the Swiss nFADP.

Yes — that is a core design goal: Swiss-only data path, operator-blind encryption, per-document access control and a complete audit trail align with FINMA outsourcing and data-handling expectations. Banks additionally value that every answer carries citations and relevance scores.

Three ways: your data path — SovraRAG runs entirely in Switzerland instead of US clouds; your corpus — answers come from your documents with citations, not from generic training data; and your controls — per-document ACLs, audit logs and per-tenant encryption are built in.

Prices are in CHF, billed annually: Solo CHF 49/month (1 user), Team CHF 490/month (up to 10 users), Business CHF 1’490/month (up to 30 users), and dedicated Sovereign Enterprise deployments from CHF 3’900/month. A 30-day guided pilot with your own documents costs CHF 4’900, fully credited to your first year. See the pricing section for details.

Yes. The standard onboarding is a guided demo, an architecture review, then a proof-of-concept with your own corpus on Swiss infrastructure from day one. Write to sales@sovrarag.ch or use the contact form.

Only the users you allow: per-document ACLs are enforced at the retrieval layer, with department-level grouping and user-level sharing. An answer can only ever draw on documents the asking user is permitted to see.