Matt zamudio

Dropbox Dash
Reimaging the workday with an AI-powered work assistant.

Educating For the new paradigm of AI

Upon arriving at Dash as the lead -- and only -- content designer on Dropbox's New Initiatives team, there were many problems to solve.

The most urgent? To simply explain Dash, since no onboarding experience existed. Users signed up, and were thrown into the app to flounder, a problem for any app, but especially an AI-powered one.

Before building out a full flow, however, there were some easy wins I could tackle immediately.

With AI comes many questions. To quickly address some of them, I worked with the designers to create system to indicate where AI was and wasn't, and what that meant.

For this, instead of inelegantly dotting the acronym "AI" or "AI-powered" everywhere, we implemented a sparkle icon to denote AI features and suggestions, along with text to explain it.

On a deeper level, I quickly saw the need to develope a solid content strategy that would inform the user experience from behind the scenes.

This would allow cross-functional alignment (PM, Exec, Design, Eng, Research, Marketing, etc.) on the description and value props of Dash in a go-to-market context, helping each team work more efficiently by giving each of us a common vision of Dash.

For example, I created guidelines such as avoiding the use of the first person ("I" or "we"). Dash, being an AI assistant, would speak for itself and avoid being self-referential ("I'm Dash, an assistant...") where possible. Dash would instead be focused on the user ("Here's some updates for your day.."), the same way a human assitant might be.

To evangelize this idea, I wrote and shared a "manifesto" featuring my proposed content strategy with examples, and eventually received team consensus on the direction.

New Onboarding Flow

After completing early work focused on refining the UX, UI, and strategy of Dash, I turned to developing a simple and educational onboarding flow, collaborating with a designer and researcher.

Beyond simple education, our onboarding flow aimed to set users up for success once they landed in the product, so that they might immediately use the product (activation) and stick around (retention).

To achieve this, I worked with design to create an onboarding flow focused on the action of adding apps to Dash, since Dash only works with source content to draw from.

In this sense, Dash is the do-it-all work app, relying on multiple sources of information and AI to help users instantly find what they need, get summarized answers about their content, and organize and collaborate with relevant suggestions from Dash.

Driving activation with content

Our new onboarding flow increased successful sign-ups, but more work needed to be done in terms of activation.

To show users how to use the product the moment they arrived, I worked with design to create messages and empty states that inspired important actions, such as...

- Adding more of apps to Dash
- Performing a search or interacting with the AI assistant
- Creating a stack, a sharable collection of work items

To acheive this, I made creative use of in-product copy, close collaboration with visual design, and research insights.

First WEek Emails

Finally, to give users the best chance of understanding and using Dash, I collaborated with a PM, designer, some marketing folks to develop a series of emails users would receive during their first week of using Dash. Each of the emails were written by myself and their design was a collaborative effort. They increased our activation rate by some ~65%.

Designing for Failure

Among other new requirements and creative implementations of content design demanded by the new AI paradigm, one of the most important was designing for failure.

To start, this meant setting the correct expectations.

When users interacted with the "Answers" features of Dash - in which they type a question about their content and Dash generates an answer - I made sure to include the line: "This feature is experimental. Info quality may vary." Transparency, I posited (and research showed) was of vital importance if we wanted to users to trust the product, a prerequisite for continued usage.

Beyond this, I worked with designers to create thumbs-up, thumbs-down feedback system. Most importantly, though, I made sure to write in situ responses for moments when Dash simply didn't know the answer to a question.

To follow this thread further, we saw user asking Dash questions like...

"What does Dash do?"
"How does Dash work?"
"What are you?"
"What can you access?"

Curious questions, to be sure, but valid ones when interacting with an entity that claims to be an work assistant.

The problem was, Dash couldn't answer these basic questions about itself because it didn't have any source material to draw from.

To this end, I worked with engineers to draft a document explaining everything someone might want to ask about Dash. Then, engineers ingested that doc into the Large Language Model (LLM) for Dash.

Now, when users ask questions like these, instead of getting an "I don't know" response, Dash respods from the source material I wrote.

The Results

While this brief overview shows only a small fraction of the work I completed at Dropbox, it's representative of some of the ways my work as a content designer and strategist can holistically enhance not only the words in a product, but the entire product direction, taking the experience from sketchy to premium.

My work helped drive growth and activation, simplify the product UI, increase education, introduce new features, define the vision, and set a path towards product-market fit.

Something I especially learned at Dropbox is to look beyond the simple UX, and to consider business goals in relation to it. By that I mean: How do we not only create a premium product experience, but one that highlights the ways in which a given product competes - or beats - the alternatives?

My work at Dropbox could not have been as fruitful as it was if not for my collaborations with the larger team, including PMs, Designers, Engineers, UX Researches, and even executives: we regularly presented our work and received feedback from Drew Houston, CEO of Dropbox, himself.

My team's ability to look at the big picture, set a standard for excellent design, and to continually push the limits is what allowed us to succeed. This was a project in which thriving in ambiguity was a requirement, and this is where I, and my team, seemed to feel most at home.

This kind of creative and forward-thinking mindset is the primary differentiator when it comes to a writer, and a design team, that either sticks to the status quo, or blazes bravely into the unknown. It's the latter trait that renders, I think, the best products in the world.