In a tiny, nearly hidden space on the second floor of the AT&T Foundry offices in Plano, some of the company’s most important innovation and collaboration work is being done.

It’s here where the future of the company is at stake, where its big bet on becoming a Silicon Valley-style fountain of high-tech ideas takes shape. It’s here, in a small, door-less nook, where the three ingredients for systemic change are housed: a small stand-up table, an assortment of Legos bricks, and a bunch of Play-Doh. 

This space—called the “ideation station,” because terrible naming conventions die hard even in the most progressive corners of corporate America—is where magic happens. It’s where the distillation of every fascinating aspect of AT&T’s nascent Foundry program (five foundries in four cities, the first one just three years old) takes place. It’s where semi-autonomous satellite groups within a huge corporate behemoth simultaneously make available the company’s vast resources to corporate partners and tiny two-man startups, while also infecting AT&T itself with Silicon Valley-style speed and innovation. In short, it’s where the idea that a Fortune 50 company can reinvent itself on the fly is tested.

“It’s where we get to play,” says Craig Lee, director of operations at one of two AT&T foundries on West Plano Parkway, the “machine-to-machine” foundry, upstairs. Machine-to-machine, which is still building out its space, helps innovators figure out how to quickly build, scale, and ship products. Downstairs is the company’s first foundry, started in 2011, which figures out enterprise software solutions. 

Lee is at once joking, and not. Even in this loose environment upstairs, where every huddle room is named after a classic video game, there’s a singularity of purpose that underscores every lighthearted touch: namely, that the foundries’ missions are to greatly accelerate innovation within the company and with outside clients. In the case of the machine-to-machine foundry, this means that, instead of figuring out how to build, scale, and ship a new product within six to 12 months, they want to do it in six to 12 weeks.

So, one of the first things the team will do once it starts a task—literally, like within an hour of meeting everyone involved in the project—is head to the nook with the Play-Doh and Legos to fashion a model. This gives them something concrete and three-dimensional to which they can aspire, something physical that will a) solve the problem before them, and b) give them a sample of just what that product will look like. So if you want to, say, build a device that can be installed on your luggage to track it on your phone, and even trigger a blue light on the luggage when it hits the airport conveyor belt, you’ve got an hour to fashion one out of the blob of reusable modeling compound before you. Everything you will do for the next few months starts there.

Lee shows off the typical routine for a new client: from Play-Doh to another room with whiteboards to meeting with the half-dozen or so other experts on staff to consider all the components and FCC regulations and scalable concerns your product will need. “So that, within a day or two,” he says, “we have a functioning Frankenstein of your product, a mock-up we can get to work on.” Within weeks, ideally, the team then has a “proof of concept,” often a 3D-printed working prototype of a device it can take back to the company, whether that’s two people in a garage or a Fortune 500 partner, to seek approval and get it to market. Easy, right?

No, says anyone who has ever tried to size a huge corporation for its innovation glass slipper. By their very nature, brands like AT&T (or any large corporation) are designed to be like warships instead of PT boats, slow to turn but able to withstand or deliver mighty blows. But even a company as large and successful as the Dallas-based telecom giant realizes just how much the landscape of high tech has changed.

That’s why AT&T set out to create a satellite center that could act like a small software company, that could produce rapid-fire solutions to industry challenges before their competitors. Not that its rivals are exactly standing still. Verizon and Sprint, for example, both have their own open-source, Silicon Valley-style approaches to innovation.

One advantage AT&T had in this game, though, was the breadth of its resources. With annual revenue upwards of $127 billion, it didn’t need to wow nouveau riche venture capitalists with buzzwords and made-up revenue projections to fund its research. In fact, together with its partners, the company has pumped $100 million into the Foundry program so far. But AT&T’s size was also its weakness: initially, few sought out the company as an innovation partner because the company’s brand identity was one of a stodgy, inflexible behemoth, too large and stubborn to give the venture the room it needed to breathe.

Three years later, AT&T has proven the skeptics—even those internally—wrong. Its five foundries—the two here in Plano, plus one each in Palo Alto, Calif.; Atlanta; and Israel—are not only finding new ways to develop new products and hurry them to market. They’re also helping reform-minded managers make AT&T a more nimble, more modern giant. “We are a change agent for the company,” says Mike Berry, the director of the first foundry, the enterprise-software-solutions downstairs one in Plano. “We’ve established a model that works.” 

One that not only works for AT&T, but one that also has shown thousands of other companies a path to cultural change for their businesses. Some have worked with the small teams of foundry engineers and been able to take back best practices and products they can use to make their businesses more adaptable to change. Others have seen how the nimbleness of these five teams has affected the way AT&T’s roughly 247,000 other employees get their work done as well.