Local AI runs a model on your own device. Start with one simple task and one app; you can explore larger models and agents later.
1. Pick a small, useful task
Choose something you can judge yourself: rewrite a paragraph, brainstorm a few ideas or summarise a short piece of non-sensitive text. Keep the original nearby so you can spot mistakes. Find your first project →
2. Choose the app before the model
The app is the interface you use. The model is the engine it loads. An approachable desktop app can handle downloading and starting a model without making you assemble a toolchain. Explore the app guide →
3. Check your computer
Look up your operating system, available memory and free storage. Start with a small model supported by the app. Leave room for the operating system and your other apps; a model download size is not the complete memory requirement.
4. Check where the work happens
A local app can also offer cloud models or online tools. Check the selected model and any connected services before adding private files. Understand local, cloud and hybrid →
5. Keep a tiny test notebook
Save three prompts you actually use. Record the app and model version, how long a response feels, and whether the answer is useful. Your own repeated tasks are more informative than a leaderboard alone.