The Hong Kong Telegraph - Dropbox CPO Mike Torres on Using AI to Reduce Workflow Friction

Hong Kong -

IN THE NEWS

Dropbox CPO Mike Torres on Using AI to Reduce Workflow Friction
Dropbox CPO Mike Torres on Using AI to Reduce Workflow Friction

Dropbox CPO Mike Torres on Using AI to Reduce Workflow Friction

Mike Torres, Chief Product Officer at Dropbox, argues that the true value of artificial intelligence in business lies not in isolated technological feats but in its ability to reduce administrative friction within established workflows. By connecting disparate information sources and automating repetitive tasks with human oversight, companies can help employees focus on high-value decision-making rather than file management.

Text size:

As artificial intelligence becomes deeply embedded in corporate infrastructure, the primary challenge for many organizations has shifted from adoption to realization. The focus is now on extracting tangible benefits from these technologies, specifically regarding efficiency and effectiveness for the workforce.

Mike Torres, Chief Product Officer at Dropbox, identifies a critical bottleneck in this transition: the gap between possessing information and being able to utilize it effectively. While businesses have accumulated vast amounts of content over decades, that data is often fragmented across various tools, folders, and systems. The process of locating the correct document, understanding its context, verifying it is the latest version, and integrating it into a workflow consumes significant time.

Torres notes that while individual interruptions may seem minor, their cumulative effect across a team is substantial. Employees spend considerable time managing information—searching, moving files, checking versions, and repeating administrative steps—rather than using it. This friction complicates collaboration, as team members frequently lack shared context or access to the necessary data.

The issue is not necessarily a lack of technology but rather the inability to make existing organizational information more accessible and useful. Many enterprise tools were designed to solve specific, isolated problems such as file storage, internal communication, project management, or information sharing. While these tools have streamlined individual activities, work rarely occurs within a single platform.

A typical project may involve documents in one system, conversations in another, feedback in a third, and the final output in yet another application. The information exists, but the contextual links between them are often broken. Employees might find a document without knowing the rationale behind a decision, whether it is current, or how it relates to other project components.

According to Torres, simply adding another tool does not resolve this fragmentation. The opportunity lies in connecting information and context across the work employees are already performing. AI facilitates this by helping users understand and interact with information in various formats, reducing the need for perfect organization or centralized storage.

There is a distinct difference between AI being impressive in isolation and being useful as part of a business process. While asking a model to summarize a document or generate a first draft can be valuable, it represents only one step in completing a task. The more significant challenge is connecting AI to the specific information and context a person needs and ensuring the output can be seamlessly integrated into the next stage of their work.

This requires looking beyond the AI model itself to consider where relevant information resides, what permissions apply, how outputs are reviewed, and how they fit into existing workflows. The most useful applications are those that remove friction from familiar processes.

At Dropbox, this principle has informed the development of a new experience designed to help users move from finding and understanding information to advancing projects using deep context from their existing files. Torres emphasizes that customer choice remains vital. Users should have the freedom to utilize AI tools that best suit their needs while keeping their content connected and under their control.

Employees can work with their Dropbox files in supported external tools such as ChatGPT, Claude, Copilot, and Muse, with outputs saved back to Dropbox. This provides a durable home for both original content and resulting work, allowing teams to review, refine, and share materials as tools and needs evolve.

Torres advises that businesses should start by identifying the specific problems they aim to solve. AI earns its place when it makes work easier and helps people carry it through to a useful outcome. A sensible starting point is assigning agents well-defined, repetitive tasks where the desired outcome is clear and the work can be verified.

Processes involving organizing information, moving between files, or executing routine steps are prime candidates for automation. However, giving an agent the ability to take action differs from asking it to provide information. Businesses must carefully consider what an agent can access, what it is allowed to change, and when human involvement is required.

Existing permissions should remain in place, actions should be visible, and people should have the ability to review and reverse changes where possible. This principle underpins Dropbox’s new experience, which identifies relevant content, proposes a plan before making changes, and requires user approval. It operates within existing permissions and provides a record of actions taken, with the ability to undo supported changes.

The goal is not to remove people from the process entirely but to delegate work that does not require constant manual attention while keeping humans responsible for decisions, judgment, and oversight.

Early examples demonstrate potential gains in time savings on repetitive tasks. At Intelligent Communities Group, CEO Randall Barney used the new Dropbox experience to organize 1,200 images into folders in approximately ten minutes and add each image’s resolution to its filename.

Creative agency YYES utilized the same experience to reorganize a project folder in two or three minutes, a task that typically takes 15 to 20 minutes. For these customers, the value was not just speed but transparency; the system showed intended actions and requested approval before execution. Similarly, Let’s Cook Video Editing Services connected Dropbox with Claude to reduce a project folder setup task from around 15 minutes to about one minute.

These examples highlight a common thread: people were previously spending time on administrative work before engaging in core tasks. AI is helping reduce those steps while allowing individuals to define goals, provide direction, and retain oversight.

Torres suggests that businesses begin by understanding how work is currently performed, including the time it takes, manual steps involved, and where delays or rework occur. This provides a baseline for assessing whether AI is genuinely improving processes.

Speed is only part of the equation; quality and workflow continuity are equally important. A draft produced in seconds has limited value if it requires hours of subsequent checking or preparation. The more useful measure is whether AI helps teams reach finished outcomes with less friction while maintaining necessary standards.

Reducing time spent on organizing files or setting up project folders can make a practical difference to daily operations. As teams build on these improvements, they can explore wider benefits such as smoother handovers, less rework, and more time for client work or tasks requiring judgment.

Torres concludes that listening to the people doing the work is as important as analyzing numbers. Their experience reveals where AI is making a difference and where processes still need attention. Ultimately, success is defined by helping people complete useful work and making it easier for others to build upon it.

宋-H.Sòng--THT-士蔑報