Eyeing Office? OpenAI is reported to be developing a ChatGPT agent feature, challenging Microsoft's dominance in office software

Microsoft's own artificial intelligence (AI) assistant Copilot has already been far surpassed by ChatGPT, and OpenAI has been reported to be challenging Microsoft's dominance in office software by developing a ChatGPT agent feature for office applications.

On Wednesday, July 16, Eastern Time, media reports indicated that OpenAI is developing a new AI agent feature for ChatGPT that allows users to create and edit spreadsheets and presentations compatible with Excel and PowerPoint directly within the chatbot, without using Microsoft's native applications. This move could reduce ChatGPT users' reliance on productivity tools produced by companies like Microsoft, marking an important shift for AI assistants towards becoming fully functional work platforms.

Reports cited sources familiar with the product as saying that the new feature will add dedicated buttons below the ChatGPT search bar to guide users in creating spreadsheets or presentations. Although the release date has not been confirmed, it indicates that users will be able to download and use various non-Microsoft applications to open these Excel and PowerPoint files.

These productivity features will make ChatGPT a more attractive business tool, posing a threat to the productivity suites sold by companies like Microsoft and Google. Ironically, due to the financial arrangements between the two companies, ChatGPT runs almost entirely on Microsoft servers.

New Agent Features Will Expand ChatGPT's Work Capabilities

The new agent features being developed by OpenAI are not limited to spreadsheets and presentations. These AI agents will be able to handle multi-step tasks without human guidance throughout the process, significantly expanding ChatGPT's capabilities.

It is understood that the upcoming agent features will also assist users in generating reports based on corporate or public data, or handling repetitive tasks involving websites, such as scheduling and booking meetings. These features are designed to save users time, similar to having interns handle mundane tasks.

Additionally, OpenAI has developed collaborative tools that allow multiple ChatGPT users to chat or collaborate on documents within the application, although this feature has not yet been released. This functionality is core to traditional productivity applications, indicating that OpenAI is attempting to establish a comprehensive work platform.

Technical Implementation Does Not Require Microsoft Authorization

OpenAI's ability to create files compatible with Excel and PowerPoint is due to Microsoft having open-sourced these file formats. This means OpenAI does not need Microsoft's permission to develop these features.

The new spreadsheet and presentation agents represent a whole new capability for ChatGPT. Users can generate and edit Excel spreadsheets and PowerPoint presentations using natural language prompts, such as "build a discounted cash flow analysis for Apple using publicly available financial documents," or "create a presentation showing how TSMC's capital expenditure costs have changed over time." According to individuals who have used these features, these agents generate files compatible with PowerPoint and Excel by writing code. This approach allows ChatGPT to create professional-level documents without requiring users to directly operate Microsoft applications.

Connection and Enhancement with Existing Features

The new ChatGPT report generation or web browsing capabilities are similar to the existing Deep Research and Operator features but are more powerful. The upcoming Reports agent expands the functionality of Deep Research by interacting with web pages, allowing users to click on links within articles to read other related stories and search online databases.

Compared to Operator, the new Actions agent is less likely to get stuck and is better at trying different methods when encountering issues such as blocked websites. These improvements make the agent functionality more reliable and practical.

User Experience Challenges Still Exist

Despite these powerful new features, ChatGPT's agent functionality still faces some user experience challenges. According to individuals who have used these features, the new ChatGPT agents can sometimes be slow or buggy. Tasks that a human might complete in 10 to 15 minutes could take half an hour or longer for the agent to finish.

Additionally, ChatGPT's product may not be suitable for all users. For example, it lacks a file storage system like Google Workspace or Microsoft Office for organizing files into folders. These limitations may hinder its widespread adoption in certain business environments.

OpenAI and Microsoft's spokespersons did not comment on this