Build your AI applications, hosted in France
A LLM application development platform powered by Dify: agentic workflows, RAG pipelines, and agents, installed and maintained by DINAO. Connect your local models for truly sovereign AI.
What is Dify?
Dify is an open-source LLM application development platform, production-ready. It brings together in a visual interface the design of agentic workflows (canvas), a complete RAG pipeline (PDF, presentations, and document ingestion, followed by contextual retrieval), agent capabilities (Function Calling / ReAct with over 50 integrated tools such as web search or image generation), and a LLMOps observability layer.
Every application created in Dify is immediately exposed via an REST API (Backend-as-a-Service approach), facilitating integration into your internal tools, sites, and automations. The backend is written in Python, the frontend in TypeScript / Next.js, and deployment relies on a Docker stack (API, worker, web, database, Redis, vector database).
Dify connects to hundreds of models, both proprietary and open-source: OpenAI, Anthropic, Mistral, Azure… as well as local and self-hosted models via Ollama or any OpenAI-compatible endpoint. This makes it an excellent foundation for building enterprise AI while retaining control over where data flows.
Host Dify at DINAO
Resource tiers compatible with Dify prerequisites (minimum 2 vCPU / 4 Go / 20 Go). Hosted in France, fully managed.
- 2 dedicated vCPU
- 4 Go RAM
- 40 GB NVMe
- Daily backups
- Managed & monitored by DINAO
- 4 dedicated vCPU
- 8 Go RAM
- 80 GB NVMe
- Daily backups
- Managed & monitored by DINAO
- 8 dedicated vCPU
- 16 Go RAM
- 160 GB NVMe
- Daily backups
- Managed & monitored by DINAO
1 hidden tier(s) (insufficient resources for this app) : Découverte
This application uses AI
The container hosts the application, not the AI engine (which requires dedicated GPUs) : inference is handled externally, with your own provider key. Prioritize a sovereign engine — Mistral AI, NumSpot, Scaleway, or OVHcloud AI Endpoints (France, GDPR) ; an international provider (OpenAI, Anthropic…) only if a specific capability requires it. Inference subscriptions are not included in hosting.
Technical details
You might be wondering…
Is Dify an AI, or a tool to create AI?
Dify is an orchestration platform: it does not contain a model; it allows you to build AI applications (chatbots, agents, RAG, automations) by connecting to models, whether local or external.
Does my data go to OpenAI or Anthropic?
It depends on your configuration. By connecting Dify to a local model hosted with DINAO, your prompts and documents remain in France. By pointing to an external API, it is convenient but your content goes to the provider.
Can I use my own models?
Yes. Dify connects to Ollama and any OpenAI-compatible endpoint (vLLM…), allowing you to point to internal infrastructure. You can also combine multiple models as needed.
Are technical skills required?
No for design: the visual canvas allows creating workflows and agents without coding. DINAO handles installation, the Docker stack, security, and updates. A technical team can leverage the API and SSO.
Can I change plans or export my work?
Yes. You can move up or down tiers at any time, and your applications, datasets, and configurations remain exportable — no proprietary lock-in.
