The question arises at home
The child accesses it from a device connected to the home network.
AI without internet
Famil·IA works locally: the model, guidelines and conversations are managed on a home computer, without depending on a cloud AI service to respond.
Many tools send the question to external servers, generate the answer there, and return it to the device. A local AI does that work on a computer that is physically inside the home. The child uses the application from a device at home and the conversation travels only over the home network to the computer running the model.
In Famil·IA, the prototype uses Ollama to run the local model, Node.js to coordinate the application, and SQLite to preserve the necessary information in the environment itself. The website you are reading is independent: it presents the project, but it cannot access that team or its conversations.
The child accesses it from a device connected to the home network.
The home computer generates the response without sending it to an external AI provider.
The age, the mode and the configured rules guide the way to respond.
Prototype data is maintained within the home infrastructure.
A child's questions can reveal interests, school doubts, emotions or family situations. Keeping them inside the house reduces the need to share that context with external services. It also helps explain privacy in a concrete way: where the question is processed matters.
This approach does not turn privacy into an abstract promise. The architecture separates the domestic prototype from the public web and does not open tunnels, ports or remote access for the demonstration.
The prototype has profiles, reviewable memory, local library, family configuration and twelve use spaces. Among them there is help with homework, reading, English, mathematics, programming, curiosity, calm and creation. The answer still depends on the capability of the installed model, but the entire walkthrough can be done offline with an external API.
It is online to explain the project. Your demo contains predefined responses and does not connect to Ollama or the home server.
It is installed and run inside the home. It is the application that processes conversations and applies family settings.
Yes. Once the local model is installed and prepared, Famil·IA can work without depending on the internet to generate responses.
It is not the same piece. The website is published to present the project; local capacity belongs to the domestic prototype.
No. The prototype is designed to run the model locally and not send conversations to external providers.
Continue exploring