This page is machine-translated from French. Read the French original.

Facility

WivenLLM can be installed in three ways. The application is the same in all three cases; what changes is who can use it and who exploits it.

Choose your mode

THE multi-user mode This is what unlocks roles, invitations, organizations, guest access, web widgets, and API keys. A workstation installation remains single-user by design.

Workstation

A native application for Windows, macOS and Linux. It includes everything: interface, server, document extraction, database, vector index.

1. Download the application for your operating system.
2. Install it like any other application.
3. Launch it: the initial setup wizard will open.

No network configuration is required, and no ports need to be opened. The data is stored in your user profile.

Material prerequisites. The application runs on a standard office machine if you use an external template provider. To run a template locally, You'll need at least 16 GB of RAM; a dedicated graphics card drastically improves speed. → Choosing your models

Server — Containerized Deployment

This is the recommended mode for an organization. It relies on a single container image and a configuration file.

Principle

• A single container carries the entire application.
• A persistent volume It contains the database, documents, indexes, and downloaded templates. This is the only element that needs to be backed up.
• The service listens on an HTTP port, which you usually place behind a reverse proxy providing TLS.

Steps

1. Prepare the storage volume on the host and give it the necessary write permissions for the container.
2. Create the configuration file Based on the provided template. Three values are mandatory and must be long, random secrets specific to your installation: - the session signing key, - the internal signing key and salt. Never reuse the example values. → Environmental variables
3. Start the container by increasing the volume and publishing the port.
4. Open the application in a browser and follow the initial setup wizard.
5. Enable multi-user mode Since Réglages → Sécurité, then create accounts or send invitations.

Production deployment

A few points to address before granting access:

• TLS. Place a reverse proxy in front of the service, or enable HTTPS directly at the application level with your certificates.
• Backup. Back up the storage volume and configuration file. Testing them by restoring is the only way to know they are working.
• Update. An upgrade involves retrieving the new image and restarting the container while preserving the volume. Back up your data beforehand.
• Sizing. If the models run on the same machine, they determine the memory and GPU requirements — not the application.

Server — native deployment

A containerless installation is possible, particularly for directly utilizing a GPU or integrating into an existing deployment pipeline. It assumes:

• a recent Node.js runtime environment; ;
• the installation of the three services (compiled interface, server, extraction service); ;
• a service manager to keep them running; ;
• a reverse proxy for TLS.

This method requires more operation than the container; reserve it for cases where it provides a concrete benefit.

Hosted instance

A dedicated instance is provisioned for your organization, with the same product and features. You receive a URL and an administrator account; operation, backups, and updates are guaranteed.

This is the appropriate mode when data sovereignty is required but the organization does not wish to operate the infrastructure itself.

After installation

In order:

1. First-time setup assistant — choice of template provider, embeddings engine and vector base, creation of administrator account and first workspace.
2. First steps — import a document and ask an initial question.
3. Users and roles — switching to multi-user mode and creating accounts.
4. Security and confidentiality — review of what is leaving the network, backups, access control.

Common startup problems