Artificial intelligence is moving beyond simply generating text to understanding how individuals communicate. OpenAI’s ChatGPT Work is taking another step in that direction with a writing-style feature that can learn from connected workplace applications, including Gmail, Google Drive, Slack and SharePoint.
Instead of repeatedly instructing an AI to sound professional, casual, direct or personal, the system can study examples of a user’s existing work and use those patterns when producing new content.
The significance of this development lies in personalization. Writing style is rarely defined by a simple instruction such as “make this formal.” It is often embedded in small habits: preferred expressions, sentence structure, punctuation, capitalization, vocabulary, greetings and even the way someone signs off an email.
Register for the next Tekedia Mini-MBA.
Register for Tekedia AI in Business Masterclass.
Join Tekedia Capital Syndicate and co-invest in great global startups.
ChatGPT Work can use examples from connected workplace tools to identify such characteristics and carry them into subsequent writing.
That could substantially change everyday knowledge work. Consider an executive preparing a company update, a journalist drafting correspondence, a manager responding to employees or a founder communicating with investors.
Rather than receiving a generic AI-generated message that requires extensive editing, the user could receive a first draft that is already closer to their established voice. The benefit is not simply faster writing; it is reduced friction between human intention and machine-generated language.
This capability also fits into the broader evolution of ChatGPT Work. OpenAI describes Work as an agent designed for longer, more involved assignments that can research information, work across connected applications and files, and create finished documents, spreadsheets, presentations and reports.
Its connectors already allow workplace information from services such as Google Drive, SharePoint, Slack and Gmail to become available within workflows. The writing-style feature therefore represents a deeper form of workplace context.
Previously, connected applications primarily helped AI understand what an organization knew. Now, they can also help it understand how an individual communicates.
That distinction could make AI assistants considerably more useful because effective workplace communication depends on both factual context and tone.
However, personalization introduces an important question: how much of a person’s digital history should an AI use to imitate them? Emails, internal messages and documents can contain sensitive information, confidential business discussions and personal communication habits.
Users and organizations therefore need to understand which connected sources are being used, what permissions apply and how workplace administrators control access. OpenAI’s documentation emphasizes that app availability and actions depend on plans, workspace settings, permissions and administrators.
There is also a broader question about authenticity. If AI becomes increasingly capable of reproducing an employee’s distinctive voice, the boundary between assistance and authorship becomes less obvious.
A message may sound exactly like its supposed author while being largely generated by a machine. Organizations may eventually need clearer policies around disclosure, approval and accountability for AI-generated communications.
Still, the direction is clear. ChatGPT Work is evolving from a general-purpose writing assistant into a context-aware workplace collaborator. By learning from the tools people already use.
It can potentially eliminate one of the biggest weaknesses of generative AI: the generic voice that often makes machine-written communication immediately recognizable.
The future of workplace AI may therefore depend less on whether machines can write and more on whether they can understand the person, organization and context behind the writing. ChatGPT Work’s new style-learning capability is an important step toward that more personalized model of human-AI collaboration.