Made for the way you work

Voice and attachments

Bring the problem into the chat. A screenshot, file, or spoken note can be a better starting point than a long typed description.

In practice

A useful place to start

Attach a screenshot of a layout issue and describe the change with voice input. The agent can compare the image with the files it is working on.

Give concrete context

Attach images and files so the agent can inspect the material behind your request. File references help keep questions tied to the right part of your workspace.

Speak when it is faster

Dictation turns speech into text for the composer. Read and edit the words before sending, just as you would a typed message.

Keep images manageable

Image handling respects model capability and size limits. An image-size control lets you cap sent dimensions when you prefer smaller requests, while preserving saved originals.

Get started

Use the composer attachment and voice controls. Select a model with verified image support before asking it to inspect a screenshot.

Good to know

Voice transcription and image requests may use a provider and incur charges. Text-only routes cannot inspect images. Think about the contents of screenshots and files before sending them to a cloud model.

Your next step starts here.

Get Pumpkin All features