La French Tech Grand Paris
Hub France IA
At DialOnce, data security, compliance, and sovereignty are at the core of our technological approach.
Our solutions are designed and hosted to ensure our clients maintain full control over their data, in strict compliance with the General Data Protection Regulation (GDPR) and the standards set by ANSSI.
Our production and pre-production environments are hosted by Scaleway, a French provider of sovereign cloud services, certified ISO 27001 and currently undergoing SecNumCloud certification.
We thus ensure that all customer data is stored and processed exclusively in France, within highly secured zones.
The databases and Kubernetes clusters are fully operated in-house, ensuring complete traceability, total technological independence, and exclusive access reserved for DialOnce teams.
Our entire architecture benefits from high multi-site availability, end-to-end encryption (TLS, SSH, SSL, IPsec), and rigorous backup management with storage located in Europe
DialOnce natively integrates Mistral AI, a French language generation and understanding model, providing a sovereign and transparent alternative to major international LLMs.
Conversational data processing takes place exclusively in France. For processes carried out via Microsoft Azure for OpenAI users, we apply a strict zero data retention policy.
This approach ensures an AI that is controlled, compliant, and ethical, while safeguarding the confidentiality and traceability of our clients’ interactions.
At DialOnce, artificial intelligence is designed to be efficient, measured, and responsible.
Our data scientists and PhD researchers apply the principle of energy frugality in building our solutions. The goal: to mobilize the right resource, at the right time, based on the actual complexity of the request.
An orchestrating intelligence evaluates each query and performs dynamic rerouting to the most suitable engine: a large language model (LLM) when fine analysis is required, or a standard algorithmic process for simpler interactions. This tiered approach helps reduce the energy footprint while ensuring accurate and contextualized responses.
We also optimize the energy consumption associated with LLM queries through:
This frugal design reflects our conviction: innovation is not measured by the power it consumes, but by the relevance of its use. DialOnce demonstrates that AI can be high-performing, sustainable, and efficient, serving a more responsible customer relationship.