Generative artificial intelligence (AI) services are moving away from separate windows or apps, integrating directly into various applications and chat interfaces. The concept of 'seamless AI integration' is becoming a reality.
On October 13, Microsoft introduced features that link the Copilot chat interface with document editing in Word, Excel, and PowerPoint. Users can modify documents alongside the chat window and make necessary changes based on comments attached to the documents. This feature will be rolled out gradually to customers in the Frontier program.
Google is also connecting AI across its suite of workplace services. On September 9, Google unveiled a feature in Google Workspace that utilizes the Gemini AI model to perform tasks across multiple apps. Users can draft documents or organize materials without switching between apps, simply by making requests from their current screen in Gmail, Google Drive, Docs, or Slides.
Previously, users had to switch to specific apps to execute AI functions for document creation or data analysis, but enhanced app integration has reduced workflow interruptions.
Independent developers are taking note of this usability. On October 12, developer Franz released the open-source tool 'BigArrow' on GitHub. This tool works in conjunction with AI coding tools like Claude and Codex, displaying arrows on the screen to indicate buttons or menus that users need to interact with, guiding them through the next steps. Even those unfamiliar with document editing can follow the tool's guidance to navigate the editing process.
For instance, when adding slide effects in a presentation program or changing system settings, the tool sequentially highlights the necessary menus and buttons. A demonstration image shared on GitHub features the phrase, 'Mom, click PDF, then Save as PDF,' alongside an arrow pointing to the relevant button, assisting users who may not be accustomed to document tasks or program usage.
BigArrow differs from automation tools that replace clicks or keyboard inputs. The developer explained that the tool is designed to point to specific locations on the screen rather than perform clicks or inputs directly. When an AI agent reaches a stage requiring human judgment or action, it provides specific guidance. The prompts disappear once the user completes the action or after a set period.
While there are differences in implementation and scope, the commonality lies in maintaining workflow continuity without switching to separate services.
Industry insiders noted, 'Seamless AI is not a new concept but rather an enhancement of existing AI tools, allowing for use without separate processes. As the need to open a separate window for AI responses diminishes, the boundary between AI utilization and actual work execution is narrowing.'
* This article has been translated by AI.
Copyright ⓒ Aju Press All rights reserved.