Better local (and open-source) AI models will have a profound impact on how we will use software. It will bring the models straight to your device, meaning more privacy and less friction.

• More privacy will result in us trusting using AI on a lot more matters than currently sharing sensitive topics via chatbots.
• Models being available on edge results in less friction for users. We’re already seeing some local model use for summarising articles on iPhone which is on par speed-wise with ChatGPT.

When Apple Shortcuts brought LLM use to making automations, I built a simple morning report automation that activates once my phone registers I’ve woken up. There I added a news summariser based on RSS feeds of local Estonian news stations, weather report, reminders and calendar recap. Few UX quirks need to be ironed out and local LLMs lack in accuracy, however, they will definitely improve long-term with new model releases. But already I am seeing a paradigm shift happening towards using a new tool for automation.