AI leaves the separate tab and embeds itself in work

During the first phase of generative AI, using a model meant interrupting work: opening a website, copying context, requesting an output, and pasting it back into the destination application. This week demonstrates the dismantling of that pilgrimage. Google Pics is being deployed as a native Workspace application and can be opened from Docs or Slides; Adobe for Slack turns conversations, files, and Canvases into raw material to create or transform content; UiPath sells the ability to coordinate agents, robots, systems, and people in an end-to-end process.

Pics is the most visible example. It allows users to generate images, import files from Drive or Photos, and select objects to edit separately. It can reformat or translate text elements, crop for different media, scale to 2K or 4K, chain edits, and revert changes. Its business advantage is not that Google has invented visual generation, but that the asset appears where the team already shares permissions, versions, and comments. A presentation image is no longer an exported file from an isolated tool but becomes part of the collaborative document.

Adobe for Slack takes the same logic one step further. It is available for Business+ and Enterprise+ plans and incorporates more than seventy capabilities from Firefly, Express, Photoshop, Premiere, Acrobat, InDesign, Illustrator, Stock, and Lightroom. Slackbot can retrieve the context from a conversation or a Canvas and call the appropriate tool: search for Creative Cloud assets, create a PDF, adapt assets for various channels, or edit image batches. The professional can continue in the full applications when they need pixel-perfect precision, but the first draft and many variations are now born in the channel where the decision was made.