Ask ChatGPT

How it works: Runs from anywhere via something like the Action Button, Back Tap, widget, or share sheet. Gets your query (typed, spoken, or predefined text, with optional images). Asks your chosen GPT model, allowing it to automatically search the web if needed. Summarises, stores, and displays your prompt and the response, either via Notes or PDF files saved to your location of choice. OR Returns the response to the calling shortcut. OR Performs an action on your device (e.g., adding an event to your calendar). Particularly when in the highly recommended (Responses) API mode, this can actually be better than the integration with Siri / Apple Intelligence because: This can make use of your device, such as using its current location or performing actions (e.g., creating a calendar event) via function calling. Siri's dictation / speech-to-text transcription capability isn't reliable, whereas this uses OpenAI's Whisper which is very accurate. Siri's recognition that you want to supply ChatGPT with a screenshot isn't reliable, whereas this has dedicated and reliable modes. Apple Intelligence doesn’t let you pick the GPT model, whereas this does. Only the iPhone 15 and newer support Apple Intelligence, whereas this should work on anything. This allows for device-specific customisation like persistent prompt instructions. If an image can be included and isn't already: If you chose during the initial set-up to always be prompted to pick a model: If you chose to store the output in Notes: If you chose to store the output in files: