Privacy Policy

Responsible Entity

The entity responsible for data processing within the meaning of data protection law is:

Elwin Lovelace AG
Waffenplatzstrasse 54
8002 Zürich
Switzerland

Email: jaron@getelwin.ch
Website: www.getelwin.ch

Data Storage in Switzerland (Data Residency)

We place the highest value on data sovereignty. Unlike many international providers, we store and process all user data and uploaded documents exclusively on servers in Switzerland.

• Hosting: Our applications run on Swiss infrastructure operated by Infomaniak Network SA. The database, authentication, and document storage are operated by our processor Supabase in its Swiss region (data center in Zurich).
• No Foreign Transfer: Your contract data and personal data are stored and processed exclusively in Switzerland. Only for technical error monitoring do we transmit scrubbed diagnostic data without any user-identifying information to the EU-hosted service Sentry.
• Processors: We use the following service providers: Supabase (database, authentication, document storage; Switzerland), Infomaniak Network SA (servers, AI models, backups, email delivery; Switzerland), and Sentry (error diagnostics; EU).

Use of Artificial Intelligence (Privacy-First Architecture)

To provide you with the services of the app, your documents and inputs go through a multi-stage technical processing chain. We use state-of-the-art technologies such as Machine Learning and Deep Learning.

The technical process is as follows:

• OCR (Optical Character Recognition): Uploaded documents (e.g., PDF scans) are made machine-readable through text recognition.
• Classification: An algorithm analyzes the document to automatically identify the type of contract (e.g., "household insurance").
• Storage & Processing: The extracted data is stored in a structured manner in our database (in Switzerland).
• RAG & LLM (Retrieval Augmented Generation): When you ask the chatbot a question, our system searches for relevant text passages in your documents and feeds them to a Large Language Model (LLM) to formulate a precise answer.

Privacy-First Architecture:

• Self-Hosted Models: We operate these open-source AI models and algorithms exclusively on servers in Switzerland: on our own infrastructure and on the servers of our partner Infomaniak Network SA.
• No Public AIs: We do not send your data to public API interfaces of third-party providers (such as OpenAI/ChatGPT, Google Gemini, etc.).

Use for Improvement and Personalization (Profiling)

In addition to pure contract management, we use your data to improve our services and provide you with personalized added value.

• Algorithm Optimization (Machine Learning): We use anonymized or pseudonymized data points from classification to train our classifier models. The goal is to identify documents faster and more accurately in the future.
• Recommender System & Profiling: We analyze the content of your uploaded documents as well as your interactions in the chat to create a user profile. Based on this profile, we develop a recommendation system ("Recommender") that can show you future optimization potentials (e.g., coverage gaps or better policies).
• Chat Analysis: Your inputs in the chat are stored and analyzed to improve the quality of responses and better understand your needs for the recommender system.

Processing of Particularly Sensitive Personal Data

As part of using the app, you upload documents (e.g., insurance policies) that may potentially contain particularly sensitive personal data (e.g., health data in health insurance documents).

• Consent: By uploading these documents and registering, you expressly consent that Elwin Lovelace AG may process them for the purposes mentioned above (analysis, LLM processing, profiling).
• Revocation: You can revoke this consent at any time by deleting the relevant documents or your user account in the app.

Disclosure to Third Parties (No-Selling Clause)

• No Data Sales: Elwin Lovelace AG does not sell user data to third parties (no address traders).
• Partners: In the current beta phase, no data is passed on to insurance partners or brokers. Should this become possible later for specific offers (Recommender), disclosure will only occur after an explicit, separate opt-in by you on a case-by-case basis.