Flash Teams as the Future of Work: A Conversation with Dr. Melissa Valentine and Dr. Michael Bernstein

By Dr. Melissa Valentine, Dr. Michael Bernstein, Teng Liu, Allie Blaising and Gabby Burlacu 
‍

As AI innovation accelerates and organizations infuse AI tools into everything from product development to customer experience, many traditional work structures– foundational to how companies hire, develop and reward– have proven too slow for today’s rapid pace of change. One of those traditional structures is teams. 

Decades of research inform effective team development and management. This is partly because it’s so easy for things to go wrong: ineffective processes, confusing hierarchies, unclear decision-making authority, and role-based limitations have long been common experiences when working in a team. All of these create drag in a world where speed is expected to accompany quality. 

Dr. Melissa Valentine, a tenured Associate Professor in the Department of Management Science and Engineering at Stanford University, and Dr. Michael Bernstein, Professor of Computer Science at Stanford University and Bass University Fellow, both hold the title of Senior Fellow at the Stanford Institute for Human-Centered Artificial Intelligence (HAI). They recently joined us to discuss their new book presenting compelling research on flash teams: groups of (often independent) experts who come together quickly to achieve a specific objective, then dissolve when the work is complete. 

Our conversation is part of Upwork’s Reimagining Work, which is a lecture series designed to provide a forum for expert practitioners and academics to foster the exchange of views on the present and future of work. During the discussion, Drs. Valentine and Bernstein shared what their research has taught them about the creation of flash teams, and why the rapid forming-focus-disband cycles inherent in flash teams are (1) more possible than ever and (2) the right fit for the demands of modern work. 
‍

1. Gabby: How did you first become interested in studying flash teams, and what made you decide to write a book about them? 

Michael: I had spent years working on research on crowdsourcing within the computer science field. Everyone seemed to think that crowdsourcing was about micro tasks like on Amazon Mechanical Turk or Prolific. This got me incredibly frustrated, both because I felt like it was often work that was less interesting for workers than it could be, but also because I felt like it was fundamentally limiting the kind of work that could be done. So I started this research on Flash Teams as a way to prove a point that computation, and what we'd otherwise call crowdsourcing, is actually a much broader phenomenon that is about the computational mediation of all kinds of work, not just microtasks. And at the time, I started connecting with a platform—what was then called oDesk, now called Upwork—as a platform that would help me realize those ideas. 

Melissa: I first became interested in flash teams years ago as a PhD student, when I studied how emergency departments were redesigning their work structures so that clinicians who had never worked together could still coordinate effectively. When I later presented that research at Stanford, one of Michael Bernstein’s students immediately saw the connection to the coordination challenges in crowdsourcing—bringing together experts who had never met and getting high-quality work from them fast. That’s what sparked our collaboration and the early research on flash teams.

We decided to write the book because we kept meeting smart, capable people who had never heard that this was even possible. I once helped someone solve a painful software-integration problem in a couple of days using a flash team, and she said, “Why have you never told me about this?” A CEO at a leadership retreat reacted the same way: “This is amazing but I’d have no idea how to do it.”

Those moments made it clear that the capabilities we see every day at Stanford, including dynamic expertise, global talent pools, AI-supported coordination, weren’t yet accessible to most leaders. The book is our way of opening that world up and giving people a practical guide for how to use flash teams in their own organizations.
‍

2. Allie: Much of the existing teams research focuses on human performance drivers that develop over time - such as cohesion, trust, and shared identity. What does your research suggest about sustaining motivation and satisfaction when AI + human teams form and disband rapidly?

Michael: The research points us at other kinds of temporary organizations—groups like disaster response teams, ER crews, film crews—as the examples for how we organize project-based work. There are strategies that these kinds of organizations use for management, for motivation, for fair payment, all providing inspiration for how we can envision a future of work with flash teams. 

Melissa: Our research shows that in teams that form and disband quickly, motivation and satisfaction don’t come from long-term cohesion, instead they come from designing the work so people can contribute meaningfully right away. In fast-cycle settings like ER crews, data-collection pilots, and flash teams, we consistently see that motivation is sustained when three conditions are in place: rapid role clarity, small but concrete signals that the work is being done ethically and well, and immediate visibility of impact. Even modest design moves, for example clear handoffs, transparent consent language, lightweight governance structures, or quick feedback loops, can give people confidence that they know what to do, that the work is responsible, and that their contributions matter. In temporary teams, the work itself becomes the source of meaning and motivation.
‍

3. Ted:  What design principles feel most important when designing for both efficiency and belonging in AI-enhanced flash team structures?

Michael: It all comes down to the human element. If you optimize for just one of those two objectives, you're going to get something that will not work. You ultimately are looking for something that jointly maximizes both objectives. There's no magical design principle here, but ultimately you have to benchmark each system or intervention against those two objectives: How is this change impacting performance? How is this change impacting belonging? 

Melissa: I see the same need to hold both goals at once. In practice, the design choices that matter are the ones that create smoother coordination and reinforce people’s sense of agency and contribution. In AI-enhanced flash teams, that often means structuring rapid role clarity, transparent handoffs, and visible impact so that AI accelerates the work without sidelining anyone. The systems that work best reduce friction, keep expertise legible, and make it easy for people to understand how their actions fit into the whole, which is what sustains both efficiency and belonging.
‍

4. Gabby: What are the potential drawbacks of too much structure in a flash team? 

Michael: We actually wrote a whole paper about that! If you put too much structure into a Flash team, then the team can't turn on a dime. In other words, an over-structured team is doomed to follow exactly the process that you laid out for it, whether or not that's the right process. We would see predefined Flash teams continue on their track, creating more and more compounding problems for each other with no way to fix it. That's why successful Flash teams need to have the ability for the team to propose changes to their own structures, to change the tasks, the goals, the personnel, and to have a hierarchy that can review and approve those changes. Without that, you're limited to only tasks that are so simple that you can carefully predefine every single step without worrying that it will go off the rails. 

Melissa: Too much structure can lock a flash team into a workflow that no longer fits the problem. I’ve seen teams keep executing a scripted process even as the work shifted around them, simply because there was no built-in way to revise the plan. When structure becomes rigid, small mismatches compound into real coordination failures. The safeguard is to pair structure with lightweight mechanisms for proposing and approving changes to tasks, roles, or goals. In fast-moving flash teams, structure should orient the work, but there need to be tools that enable people to constantly update and realign the structure. 
‍

5. Allie: How do we balance individuals' desire for control over their reputation in a team when AI agents become part of the flash team loop?

Michael: I think that many enterprising workers are already attempting to integrate AI agents as part of their work. The smart ones are realizing that sometimes these things slow them down more than they speed them up, or are figuring out exactly what kinds of work these agents can actually accelerate. Ultimately, in every project, there is going to be someone who is responsible for the ultimate work product. And it's going to be that person whose reputation will be on the line if the AI agent makes a mistake. So ultimately, I think the biggest challenge is making sure people are calibrated in when and how to use these agents. 

Melissa: Like Michael said, the challenge is calibration. People are learning quickly where AI helps and where it introduces risk. No matter what AI contributes, responsibility still lies with the human expert, so judgment becomes critical. The opportunity is that workers who use AI thoughtfully can extend their capabilities without compromising their professional reputation.
‍

6. Ted: What open questions are you most excited about as AI begins to provide an intelligence layer to how teams form, coordinate, and evolve?

Melissa: I’m especially energized by how this moment aligns with the “AI + Organizations” Grand Challenge that we are organizing with leaders at Google DeepMind. Together, we’ve issued a call to build the basic discipline of organizational design for the age of AI.  This frame recognizes the need to move beyond individual use of AI tools toward systems that make teams and organizations collectively more intelligent. The big opportunity is to rethink how we coordinate, lead, and structure work when AI becomes an intelligence layer in organizational design. We have the chance to design human–AI teaming systems that elevate our collective intelligence and value.

Michael: At the end of the day, for me, it's most exciting to think about what kinds of collaboration and coordination we could do better than we do today. So many of our most ambitious goals are managed completely based on experience and intuition about how we ought to be working together. Could AI help us make better decisions? Could it help us foresee when there are going to be issues? I think that if it could do that, it could make our work lives happier, healthier, more fulfilling. 

To learn more about flash teams, check out Flash Teams: Leading the Future of AI-Enhanced, On-Demand Work.

‍
‍
About Dr. Melissa Valentine

Stanford Professor Melissa Valentine is a leading expert on technology and business. Her research delves into the evolving interplay between artificial intelligence, algorithms, and organizational design. Through in-depth field studies, she explores how technology reshapes workplaces, providing valuable insights into the future of work in a digital and specialized world. Her award-winning research has been featured in The New York Times, The Wall Street Journal, Harvard Business Review, Wired, Fast Company, and the Financial Times.

About Dr. Michael Bernstein

Michael Bernstein is an Associate Professor of Computer Science at Stanford University, where he is a Bass University Fellow, Senior Fellow at the Stanford Institute for Human-Centered Artificial Intelligence, and Interim Director of the Symbolic Systems Program. His research focuses on designing social, societal, and interactive technologies. His research has been reported in venues such as The New York Times, TED AI, and MIT Technology Review, and Michael himself has been recognized with an Alfred P. Sloan Fellowship, the UIST Lasting Impact Award, and the Computer History Museum's Patrick J. McGovern Tech for Humanity Prize. Michael holds a bachelor's degree in Symbolic Systems from Stanford University, as well as a master's degree and a Ph.D. in Computer Science from MIT.

About Allie Blaising

Allie Blaising is a Lead User Experience Researcher at Upwork, where she leads customer research that shapes design and business decisions across multiple verticals, with a recent focus on Small Business Growth and Generative AI product initiatives. She holds a Master’s in Science from University of Pennsylvania.