AI in the Workplace: Who Really Goes the Extra Mile?

AI in the workplace accelerates the beginning, not the end

Artificial intelligence is rapidly lowering the barriers to producing initial results. However, simply using AI in the workplace does not automatically mean that the work is complete. A study commissioned by Dropbox highlights the gap between where work begins today and the effort required to actually complete it. While AI can generate an answer in seconds, turning that output into a final result often requires moving content between applications, sharing it with colleagues, and preparing it for approval. 84 percent of respondents agree with the statement that AI helps employees create content faster than their organizations’ processes and workflows can keep up with. This highlights a new challenge in digital transformation and an unexpected bottleneck: While AI accelerates the creative process, the real challenge lies in transforming its output into trustworthy, usable, and lasting results.

Content moves through multiple tools and processes

The study’s findings show that there are still numerous manual steps involved between AI-generated results and the finished work. 74 percent of respondents report manually moving AI-generated content between different applications at least three times on an average workday using AI. 30 percent do this as many as six times or more. In addition, 65 percent of respondents use three or more applications to complete a typical AI-assisted task.

More than half of those surveyed (54 percent) now say they use an AI assistant as their primary starting point for new knowledge work. This puts AI ahead of documents, emails, search engines, and project management tools. However, only 11 percent of respondents say that AI already delivers finished work products today without any further hurdles.

Participants cited the accuracy of the results (40 percent), security requirements or company policies (26 percent), as well as additional friction caused by shifting tasks between applications, approval processes, and a lack of business context.

The New «Last Mile» of Work

Generating an initial result is just the beginning. While AI can generate this result in a matter of seconds, the subsequent processing represents the new «last mile» of AI: the journey through information, people, decisions, authorizations, and workflows required to turn it into a trustworthy and usable work product.

AI-generated work is also increasingly becoming part of an organization’s long-term knowledge base. Forty-three percent of respondents report that the majority of their weekly AI-supported work ends up permanently in team documentation or shared knowledge databases.

Content that moves around needs a permanent home

According to Dropbox, to overcome the challenges of this “last mile,” it is not enough to focus solely on content generation. As work shifts between humans and AI, the associated content and contextual information must also transition seamlessly. At the same time, the resulting outputs need a permanent home where teams can further develop and build upon them.

Dropbox aims to support this by linking content and context across various AI applications and providing teams with a single, unified place where they can continue their work. According to the company, usage of Dropbox’s AI partner tools has increased 13-fold over the past four months. The company says this growth indicates that customers are increasingly integrating Dropbox into their AI-powered workflows.

About the study

From July 28 to 30, 2026, Dropbox conducted a survey of 504 U.S. B2B professionals who use AI. The respondents came from various fields, including IT, engineering, sales, marketing, finance, operations, human resources, and customer support.

For more information about the study, visit the official Dropbox blog: https://blog.dropbox.com/topics/company/ai-last-mile-work-study