Startups operating in the gig economy typically use software platforms to assemble disconnected armies of freelance contractors to perform simple tasks. Think of Uber, which allows customers around the globe to summon drivers to take them places from practically any street corner, using nothing more than a phone app.
Computer engineers at Stanford are taking things a step further for more skilled workers in the knowledge economy. They’ve invented something called flash teams. Although still somewhat experimental, flash teams use software platforms to break down complex work that requires collaboration, such as engineering and web design, into modules of specific tasks that the network hands off from freelancer to freelancer, in a virtual assembly line. Once the work is complete, those teams are dissolved, but they can potentially be recombined for other types of work in the future.
The Stanford engineers’ concept was put into practice in 2014 with a software platform called Foundry, which has used flash teams of about half-a-dozen people, to design web apps and create videos. Now a New York-based website design company called B12, launched in 2015, is using flash teams as a key part of its process.
Founded by venture capitalist Nitesh Banta, the company has at its core a software platform called Orchestra. The platform, based on the research used to create Foundry, uses artificial intelligence to accomplish some of the rote tasks involved in website development. For example, it uses web crawlers to scrape up information about new customers from social media profiles or other information on the web. That information is used to complete an initial website prototype within 60 seconds, and to build what Banta calls a scaffolding for the project. Like Foundry, Orchestra breaks down the web design process into component tasks that are farmed out in modules to human workers, who are all freelance contractors.
B12 has a distributed workforce of several dozen workers including website designers, photo editors, and copywriters located in the U.S., Canada, and the U.K. When new work arrives, Orchestra sends those contractors an email with something it calls a staffbot, which contains details about the project. Workers can say yes or no to the project, for which they will be paid a previously agreed-upon rate.
Once they are assigned to the work, Orchestra sets up a dedicated Slack room for communication, and then plays umpire for the workflow, ensuring tasks are completed in order and on time. For example, the system will prompt web designers to build the various website templates in one module. Once they’ve completed their work, the system creates another module for photo editors to insert and lay out images. Copywriters will next be prompted to add in the text they’ve composed.
Along the way, Orchestra assesses project efficiency, such as the length of time it takes its human workers to complete their assigned tasks. Based on this efficiency data, it also makes determinations about which website creators work well together, with the object of pairing them together in the future.
Banta says this workflow system helps B12 delivering designs up to twice as fast and at about a tenth of the cost of many competitors with dedicated design teams.
This method of combining gig workers with AI might be part of the future of work, but it’s not without its downsides. University of Michigan management professor Jerry Davis, whose research focuses on the vanishing U.S. corporation and its impact on workers, says the gig economy works well for Uber and others because they pair one person with one well-defined task, such as picking someone up and driving them somewhere. That model begins to break down with the web of interdependencies necessary to complete more complicated tasks, particularly when strangers are involved.
“Intellectually it is a much more intriguing problem, how you hire people for interdependent work and get them to work as a team,” Davis says. He adds that one possible disadvantage of the flash team model is the potential for job stagnation, as there’s no real impetus from an organization to develop workers. Another is the danger for all manner of work to become like a Home Depot parking lot, he says, with day laborers lining up for every type of job.
Banta has thought about that too. B12 has structured into its workflow a system it calls hierarchical review. That process pairs less-experienced workers with more experienced ones, who review the work and offer critiques like mentors.
Banta says his experts also benefit because they can put their skills to work more consistently. As opposed to hunting for their own projects, Orchestra gives them access to a stable stream of work. And they’re learning ways to collaborate with workers they’ve never met and who are far afield, he says.
And that may just be the beginning. “In the old way that work was put together, you needed a corporation to organize a team of people around a common mission to accomplish work,” Banta says. “But new software can now quickly organize people with a mission to do a particular job, so the challenge is to offer new ways to use technology to offer more fulfilling work.”
Jeremy Quittner is a freelance writer in New York. His work has appeared in Business Week, Fortune, and Inc.