Confidential AI for enterprise documents.

Yes. DeepConfidential has no persistent backend. Once your session ends, all documents and data are permanently removed – no archives, caches, or storage layers. True session privacy guaranteed. Session-specific confidentiality is ensured.

Absolutely not. Your documents never influence or train the AI. All information is processed exclusively during the use of DeepConfidential and is subsequently deleted. We never use your data for model improvement or any other purpose.

Whereas conventional AI services store data and reuse it for training purposes, DeepConfidential processes information strictly on a session-based and fully isolated basis. Furthermore, DeepConfidential does not access external internet sources, thereby consistently mitigating potential security risks such as code injection or other external attack vectors.

DeepConfidential does not store data, retain histories, or reuse your documents. Your sensitive information remains confidential.

No. RAG requires storing document excerpts in searchable databases, which can create privacy risks. DeepConfidential processes only the documents uploaded by the user – no indexing, no database, no trace after the session.

Yes. It is GDPR compliant and also FADP-compliant – the Swiss Federal Act on Data Protection – which ensures the strictest privacy standards in Switzerland. DeepConfidential is hosted within the EU/EFTA on DeepCloud’s ISO 27001:2022-certified cloud solution, meeting the highest European and Swiss data protection requirements.

Yes. On-premise deployment is available for organisations requiring full control. Sensitive documents never leave your environment, meeting the highest internal security policies.

PDFs, Microsoft Word (DOCX), Excel (XLSX), plain text, and WAV audio (with automatic transcription). All formats can be processed seamlessly in a single session.

English, French, Spanish, German, Chinese, Japanese, Korean, Arabic, and major Asian & MENA languages.

We provide enterprise-grade support for all customers, with dedicated assistance for on-premise deployments. Our team is available to ensure smooth setup and ongoing assistance.

No. DeepConfidential cannot hand over your data to any government because it has nothing to hand over: it is operated under Swiss jurisdiction and, more importantly, is built so that no user data is stored by design. The system processes your input only in-memory during the active session and does not use RAG, databases, logs, or other persistent storage for your content, so once the session ends, the data is technically no longer available to be requested or disclosed.

The key difference is that NotebookLM is explicitly built as a persistent RAG notebook, while DeepConfidential is a non‑persistent, session‑only tool.

• NotebookLM lets you upload sources (PDFs, Docs, links, etc.) and builds a lasting, searchable knowledge base (RAG pipeline with vector storage) that stays available across sessions in your notebooks.
• Your uploads, notes, and saved responses are stored in Google’s infrastructure under your account and can be re‑used, re‑indexed, and combined with new prompts later.
• By contrast, DeepConfidential does not store any user data at rest: no document database, no vector store, no RAG index, and no persistent logs. All processing happens in memory during the active session and disappears afterwards, so there is nothing that can be queried or re‑used later and nothing that can be handed over.

No, neither DeepConfidential nor any products of DeepCloud are subject to the United States CLOUD Act.

Yes, the analysed data can be exported in PDF or Word format.