With AI now firmly embedded in business systems and processes for many of us, the focus for many firms now turns to realizing tangible benefits from then technology.
AI technology promises much in the way of achieiving efficiencies and effectiveness for workers across the board, but what does that look like in practice?
We spoke to Mike Torres, Chief Product Officer, Dropbox, to find out more.
- Across industries, where do you see the biggest day-to-day friction in how organizations get work done, and what is the cumulative impact on teams?
A lot of the friction comes from the gap between having information and being able to do something useful with it. Businesses have accumulated enormous amounts of content over the years, but that information is often spread across different tools, folders, and systems. Finding the right document, understanding its context, getting the latest version and then bringing it into the next stage of a process can take more time than people realize.
Individually, these can seem like small interruptions. But when they happen repeatedly across a team, they add up. People spend time managing information rather than using it: searching, moving files, checking versions and repeating administrative steps. This friction also makes collaboration harder because people don’t always have the same context or access to the information they need.
The challenge isn’t necessarily a lack of information or technology. It’s making the information organizations already have more accessible and useful, so people can spend more time on the work that needs their judgement and expertise.
- Why haven’t years of investment in digital tools and collaboration platforms resolved these problems? What makes organizational knowledge so difficult to connect and use?
Many of the tools businesses have invested in were designed to solve particular problems: storing files, communicating with colleagues, managing projects or sharing information. They’ve made those individual activities much easier, but work rarely happens in just one place.
A project might involve documents in one system, conversations in another, feedback somewhere else and the final output in a separate application. The information exists, but the context around it can become disconnected. You might find a document without knowing why a decision was made, whether it’s the latest version or how it relates to the rest of the project.
That’s one of the reasons why simply adding another tool doesn’t necessarily solve the problem. The opportunity is to connect information and context across the work people are already doing. AI makes that more possible because it can help people understand and work with information in different formats, rather than relying on everything being perfectly organised or stored in one place.
- Many businesses are experimenting with AI. What separates those experiments from AI becoming a useful part of everyday work?
There’s a difference between AI being impressive in isolation and it being useful as part of a business process. Asking a model to summarize a document or generate a first draft can be valuable, but that’s often only one step in getting something done.
The harder challenge is connecting AI to the information and context a person actually needs, then making sure the result can be used in the next stage of their work. That means thinking beyond the model itself: where the relevant information lives, what permissions apply, how the output is reviewed, and how it fits into existing ways of working.
The most useful applications will be those that remove friction from familiar processes. At Dropbox, that’s informed our approach to the new experience, helping people move from finding and understanding information to moving projects forward, using deep context from the files they already have.
Customer choice matters too. People should have the freedom to use the AI tools that work best for them while keeping their content connected and under their control. They can work with their Dropbox files in supported tools like ChatGPT, Claude, Copilot, Muse, and more, with supported outputs saved back to Dropbox. That gives both the original content and the resulting work a durable home, where teams can review, refine and share it as their tools and needs change.
Ultimately, businesses need to start with the problem they’re trying to solve. AI earns its place when it makes that work easier and helps people carry it through to a useful outcome.
- As AI agents begin taking action, what should businesses allow them to do, and what controls are needed to build confidence in their use?
I think a sensible starting point is to give agents well-defined, repetitive tasks where the desired outcome is clear and the work can be checked. There are plenty of processes involving organizing information, moving between files, or carrying out several routine steps where automation can be useful.
But giving an agent the ability to take action is different from asking it to provide information. Businesses need to think carefully about what it can access, what it is allowed to change and when a person needs to be involved. Existing permissions should remain in place, actions should be visible, and people should have a way to review and reverse changes where possible.
That’s the principle we’ve applied to the new Dropbox experience. It identifies relevant content and proposes a plan before making changes, with the user approving that plan. It also works within existing permissions and provides a record of actions taken, with the ability to undo supported changes.
The aim shouldn’t be to remove people from the process altogether. It should be to delegate the work that doesn’t need constant manual attention while keeping people responsible for decisions, judgement and oversight.
- Can you share specific customer examples where AI is already changing how a workflow gets completed? What has improved, and what still requires human input?
We’re seeing some encouraging early examples, particularly around the time people spend on repetitive work involving files and content. At Intelligent Communities Group, CEO Randall Barney used the new Dropbox experience to organise 1,200 images into folders in about 10 minutes, then add each image’s resolution to its filename.Â
Creative agency YYES also used the new Dropbox experience to reorganise a project folder in just two or three minutes when it normally would have taken 15-20 minutes. The important point for the customer wasn’t just the time saved: the system showed what it intended to do and asked for approval before making changes. And at Let’s Cook Video Editing Services, connecting Dropbox with Claude helped reduce a project folder setup task from around 15 minutes to about a minute.
These are different businesses, but the common thread is that people were spending time on administrative work before they could get to the work itself. AI is helping reduce those steps. People still define what they want to achieve, provide direction and retain oversight of the outcome. These examples are early, but they show the value of applying AI to specific, practical parts of a workflow.
- How should businesses measure whether AI is improving a complete workflow, beyond generating more content or completing individual tasks faster?
I would start by understanding how the work gets done today, including the time it takes, the manual steps involved, and where delays or rework tend to happen. That gives businesses a baseline for assessing whether AI is actually making the process better.
Speed is part of that picture, but so are quality and how easily the work moves forward. A draft produced in seconds has limited value if someone then spends hours checking it or preparing it for the next person. The more useful measure is whether AI helps the team reach a finished outcome with less friction, while maintaining the standards they need.
Our early customer examples offer useful starting points. Reducing the time spent organizing files or setting up project folders can make a practical difference to someone’s day. As teams build on those improvements, they can look at the wider benefits, such as smoother handovers, less rework, and more time for clients or work that needs their judgement.
Listening to the people doing the work matters alongside the numbers. Their experience helps show where AI is making a difference and where a process still needs attention. Ultimately, success is about helping people get useful work finished and making it easier for others to build on it.

Go to Source
Mike Moore