Prompting is becoming a conversation with the canvas
Adobe adds targeted, hands-on editing controls to its ChatGPT plugin for images, designs and PDFs.
The interesting part is not generation; it is the return of precise control.
The Story
Adobe’s new ChatGPT experience lets people move between natural-language instructions and direct manipulation. Users can select a region, ask for a targeted change, then refine hue, saturation, contrast or document text through interactive controls. ([blog.adobe.com](https://blog.adobe.com/en/publish/2026/09/29/openai-devday-2026-introducing-hands-on-editing-adobe-plugin-chatgpt?utm_source=openai))
That is a more mature model of AI-assisted making than one-shot generation. The conversation handles intent; the editor handles judgement. Crucially, the work stays in context instead of forcing a designer to restate the brief after every adjustment.
For client work, this could make exploratory rounds faster—but only if the creative brief remains stronger than the tool. Use conversational editing for range and momentum, then use hands-on controls to protect specificity, restraint and finish.
Studio Wensday Angle
The best AI workflows alternate between ambiguity and precision.