Concepts

FlowtheraSI is a small set of products that work together. Everything that touches your data runs on machines you control.

The products

Flowthera Server is a program you install on a Linux machine. It downloads AI models, runs them, and offers an API that other apps can call. Every request needs an API key. It also handles updates, backups and health checks.

Flowthera Studio is a desktop app for Windows, macOS and Linux. It can run a small model on your own computer, or connect to a server. It includes a code workspace, a model finder and a set of offline tools.

Flowthera Mobile (planned) is a phone app that connects to your server so you can chat, watch its health and approve actions.

Flowthera Account is where you manage licences, downloads, billing and team members. You do not need an account to use the free tier.

Models and runtimes

A model is the file that contains the AI’s learned knowledge. Models come in different sizes. Larger models are usually more capable, and need more memory.

A runtime is the engine that loads a model and generates text with it. The server and the desktop app use well-known open-source runtimes such as Ollama and llama.cpp.

Models are downloaded from their original publishers. The app checks each file against its published fingerprint before loading it.

Licences

There are two kinds of licence to know about:

• Model licences are set by whoever published the model. Some allow any use, some restrict commercial use. The app shows the licence before every download.
• FlowtheraSI licences unlock paid features. They are signed files that are checked on your machine, so they keep working offline.

Where data goes

By default, nothing you type, upload or generate leaves your machines. External AI providers are off until you switch one on with your own key, and requests sent that way are labelled. See How it works for the full table.