Your first chat¶

Now that you know the interface, let's see how to use it to chat with an AI model.

Creating a new conversation¶

To start a chat, click the New button at the top of the left sidebar. A popup opens that lets you select the model to use:

The popup shows the default model (the one selected in the right sidebar), but you can change it for this specific conversation. Confirm to open a new chat ready to receive your first message.

Default model

The model selected in the right sidebar becomes the default for new conversations. If you always use the same model, set it there so you don't have to choose it every time.

Sending a message¶

Write your message in the text area at the bottom and press Enter or click the send button. The model will start responding immediately, with text appearing progressively like in a normal chat.

Streaming responses

Responses appear word by word as they're generated. This lets you read the beginning of the response without waiting for it to complete.

Managing conversations¶

Each conversation is automatically saved in the left sidebar. You can:

Rename a conversation by clicking on its title (when it's open in the central area)
Resume a past conversation by selecting it from the list
Delete a conversation with the trash button

Conversations are saved on your computer in the app/data/conversations folder. Each conversation has its own subfolder containing both messages and any attachments.

Exporting a response¶

You can save a single response in different formats by clicking the download icon at the bottom right of each response:

Available formats are:

Markdown (.md): ideal if you use editors that support it
Text (.txt): the most universal format
Word (.docx): for formal documents

Changing model during conversation¶

You can change models at any time by selecting a different one from the dropdown menu in the right sidebar. The new model will continue the conversation from where you left off, but will have a different "personality" based on its characteristics.

Model behavior

Different models respond differently. Some are more creative, others more precise, others better suited for specific tasks like programming. Experiment to find the one you prefer.

The first experiment¶

Try sending a simple message to verify everything works:

Hello! Can you briefly introduce yourself?

If you receive a response, congratulations: you've just run your first local AI inference. Everything happened on your computer, without any data being sent to the internet.

Tips for effective prompts¶

To get better responses:

Be specific: instead of "tell me about history", ask "summarize the causes of World War I in 5 points"
Provide context: explain what you want to achieve and why
Request a format: specify if you want a list, a paragraph, a table
Iterate: if the first response isn't perfect, ask to modify or elaborate

Local models work better with clear instructions. In the next chapter we'll see how to download additional models and choose the one best suited to your needs.