Heavybit: Grad School for Startups Building a Software Supply Chain

James Lindenbaum, the Heroku co-founder who has spent the last year building a new startup training program called Heavybit Industries, thinks the software industry is still just as primitive as the automobile business was back in the Great Depression.

As late as 1930, Ford and other automakers remained vertically integrated, even building the ore processing plants and foundries needed to make the steel for auto bodies, Lindenbaum points out. Eventually, of course, they discovered that it was more economical to buy materials like steel, glass, and rubber from specialized suppliers. Today, car companies are essentially design and assembly shops, outsourcing much of their actual manufacturing to an enormous, global network of parts makers.

Only recently has the software industry started to evolve beyond its own era of vertical integration, Lindenbaum argues. Until players like Amazon began to offer public cloud-computing services, purveyors of Web-based consumer or business software had to manage (and sometimes even build) most of the supporting infrastructure, from the raw-metal processors and switches to data centers, networks, Web servers, and databases.

Already, the emergence of a “stack” of physical and software components that companies can buy or rent à la carte has made building an Internet startup vastly easier and cheaper. But here’s the key thing, in Lindenbaum’s mind: this change is still just beginning. A lot more of the work that goes into building and running a mobile app or Web service could be outsourced, allowing the innovators at the top of the stack to focus on creating things that customers want.

As billions more people around the world come online, “a tremendous amount of consumer and enterprise software is going to be written, creating demand for all of the stuff one step below,” such as Heroku itself, Lindenbaum says. (Heroku, in case you don’t know, is a platform for Web applications—read on for more details, as it’s integral to the Heavybit story.) And someone will need to meet that demand. “I think the collective consciousness is just at the point of becoming aware that this stuff one layer down powers everything else—it provides the raw materials and tooling and allows everything else to be built.”

The Heavybit Industries building at 325 9th Street in San Francisco.

The Heavybit Industries building at 325 9th Street in San Francisco.

That’s the backstory for Heavybit, which Lindenbaum started with managing director Tom Drummond and “vibes manager” Jason Harper. They and the advisors assembling for the operation—including the other two members of Heroku’s founding team, Orion Henry and Adam Wiggins—envision it as a training ground for seed-stage startups building new products that make life easier for application developers.

In a few respects, Heavybit resembles a classic startup accelerator like 500 Startups, Techstars, or Y Combinator. It’s got a competitive admissions process, a set term of residence (nine months), a large network of mentors, a training curriculum, and a workspace for members—a meticulously renovated furniture warehouse on 9th Street in San Francisco’s SoMa district. And it asks admitted startups for a small percentage of equity, meaning, at bottom, it’s an investing operation.

But in other ways, Heavybit is a new animal on the tech savannah. It only admits companies that already have seed funding. It doesn’t provide monetary stipends or investments to startups. And the curriculum skips over “startup 101” topics like fundraising or customer development, going straight to advanced topics like the intricacies of marketing to developers. “We aren’t trying to be an incubator,” Lindenbaum says. “We aren’t trying to help [startups] figure out what they want to be when they grow up. We think of it as grad school to Y Combinator’s undergrad.”

When it comes to the accelerator experience, Lindenbaum knows whereof he speaks. Heroku went through a major pivot as part of Y Combinator back in 2008. A couple of years later, it provided one of YC’s best exits to date, when it was acquired by Salesforce.com for around a quarter of a billion dollars.

Knowing a bit about Heroku is the first step to understanding what Lindenbaum is really trying to do at Heavybit. Often described as a “cloud application platform” or a “Platform-as-a-Service,” Heroku is a place where developers can upload Web-based applications written in languages like Ruby, Java, Node.js, or Python, and count on the company’s systems to run the code and make the apps available to end users, so that its customers don’t’ have to maintain the needed servers and support software. (Technically, the Platform-as-a-Service label is inaccurate, since in practice, since Heroku outsources the hosting part of its job to Amazon Web Services; also, it doesn’t provide all the tools developers need. “It’s really a ‘Runtime-as-a-Service,’” Lindenbaum says.)

Lindenbaum left Salesforce.com early this year, and Wiggins departed in May. (Henry is still there.) But it was being inside Salesforce.com, and watching as their platform grew to support thousands of Web companies running millions of applications, that got the Heroku founders thinking in a larger way about the state of the developer-tools business.

“Because of the sheer size of the platform and the visibility we had into what developers were doing with it, it was super obvious what application architectures were becoming popular,” Lindenbaum says. It was clear, for example, that most Web companies were busy tying their applications into dozens of other tools and services, such as Amazon’s S3 for storage or NewRelic for performance monitoring. They’d become dependent on these outside services in the same way that a bike-wheel hub is supported by dozens of spokes.

“So with Heavybit, there is something clear from where we sit that a lot of people haven’t thought about yet,” Lindenbaum says. That thing is the value of opaqueness in the supply chain.

Yes, we live in an age that celebrates transparency in all things, but it’s not always what we need most, Lindenbaum argues. To return to the automotive analogy: If you’re the first person to make a car, you have to do everything yourself. But as the market demands more cars, eventually outside companies will line up to supply you with steel, glass, and car parts. “As the market starts to mature, you get this true stratification,” says Lindenbaum, who admits to being a car buff. “A car manufacturer no longer needs to think about where his steel is coming from; he just says what type, and it shows up. The opaqueness of the supply chain is very important.”

When a supply chain is too transparent, “You are seeing into it and worrying and thinking and trying to innovate on the supplier’s behalf,” Lindenbaum says. “Whereas when it’s opaque, all you have to think about is your part of the problem”—that is, understanding your customers’ needs and coming up with creative new features that they’ll be happy to pay for.

Timber, concrete, and steel give the Heavybit clubhouse a utilitarian feel.

Timber, concrete, and steel give the Heavybit clubhouse a utilitarian feel.

Only in the last five to seven years or so has a stratified supply chain begun to emerge in the Web application business. “Amazon was the first truly, philosophically opaque provider,” Lindenbaum says. With Amazon’s Elastic Compute Cloud (EC2), Internet companies no longer had to worry about building their own data centers, or even renting virtual private servers from hosting services. Amazon takes care of all the basic managerial concerns, like load balancing and disaster recovery. Heroku followed the same model, Lindenbaum says. “You shouldn’t have to care how we do it, you can just focus on building your organization.”

Marc Andreessen famously argued that software is eating the world and that more and more business processes and consumer services will turn into cloud applications delivered via browsers or apps. If you believe that, it follows that the customer-facing companies at the top of the supply chain are going to need more help from the tool and platform providers below them. Lindenbaum thinks Silicon Valley venture firms know this—it’s the main reason they’ve switched their investing focus from the consumer Internet to enterprise and B2B software. And from his study of the way stratified markets evolve, Lindenbaum says he’s convinced there’s lots of room for new businesses that want to make tools, platforms, products, and services for developers. “My argument is that we are about to witness an unbelievable explosion in the number of companies doing this kind of thing and how much value they are providing to the world. It’s not a prediction, but an unavoidable consequence of where we are.”

Heavybit's ground floor includes informal work areas and kitchen/cafeteria; the platform around the stairs serves as an informal dais for talks and events.

Heavybit’s ground floor includes informal work areas and kitchen/cafeteria; the platform around the stairs serves as an informal dais for talks and events.

The paradox right now, though, is that a lot of developer-oriented companies are struggling. “Because Heroku was very visible, a huge percentage of these guys starting these companies would come talk to us and ask for advice. And over time it started to become clear that there is a unique set of challenges that come with being one of these companies,” Lindenbaum says. “A lot of things are not figured out yet. Customers haven’t figured what they want to pay for, or how much. Going to market is hard. Even this bifurcation between the people who are using the tools—developers—and the people who pay for them—companies—is really tricky to navigate.”

Even fundraising is a challenge, despite the venture industry’s general openness to investing in B2B services. “They go out trying to raise their first round of money and they come back very crestfallen, thinking that the venture guys aren’t smart because they ask terrible questions, when a lot of the time it turns out they are just talking to the wrong investors.” Worst of all, Lindenbaum says, these entrepreneurs have no one to talk to about their problems. “Because there are still so many consumer-facing app companies, the developer companies are way outnumbered, so they feel very alone.”

Eventually, Lindenbaum says he got tired of fielding the same questions over and over and playing matchmaker between companies in crisis and experienced entrepreneurs who might have solutions. What entrepreneurs building developer tools needed, he decided, was some systematic camaraderie—a place to share best practices, learn from veterans, and help each other out. And that’s what Heavybit is designed to be, while (in theory) profiting over time from its equity stake in the companies brought into the club.

Since opening its doors eight months ago, Heavybit has admitted 15 companies to its program. Nine of them are currently working from the Heavybit clubhouse, which offers 70 desks spread across three floors. (There are also kitchen and conference areas, a video room, and places to lounge, all in a masculine concrete-and-timber setting that’s already winning architectural awards.)

The typical Heavybit company has already raised a seed round of at least $1 million and has three to five employees, Lindenbaum says. In many cases they’ve already been through an “undergraduate” accelerator like Y Combinator. Companies are admitted on a rolling basis, whenever their need is greatest, and they typically double in size during their residency. So at any given time there’s a mix of three-person and 10-person companies in the building.

Heavybit helps members with hiring and public relations, and arranges access to a large and impressive group of advisors, some of whom work from the clubhouse. Most advisors have been through one or more business cycles in the Silicon Valley software world, and are able to help entrepreneurs work through common challenges, Lindenbaum says.

The formal curriculum is built around guest speakers who cover “really specific, meaty problems that companies making developer products have,” he says. “We are trying to focus on content not available anywhere else.” Past talks have focused on how to convert developer engagement into sales; how to market to developers, and how that’s different from other types of product marketing; and how to design interfaces for developers. (“DX,” or developer experience design, is an important, emerging corner of user experience design, Lindenbaum says.)

The most famous company to participate in Heavybit so far is Stripe, which gives developers at other companies an easy way to build credit-card payment forms into their sites. Lesser-known companies include Kodowa, which is building a popular new development environment called Light Table, and Meteor, which is creating a new kind of framework to help developers write JavaScript-based Web applications faster. (Meteor has already grown so large that it had to move out of the clubhouse, Lindenbaum says.)

Heavybit's conference room features propellers from vintage aircraft.

Heavybit’s conference room features propellers from vintage aircraft.

The money to renovate and furnish the clubhouse, keep the lights on, and take care of the member companies comes from a group of prominent venture, and corporate investors recruited by Lindenbaum, including Accel, Baseline Ventures, Data Collective, Harrison Metal, IA Ventures, Ignition, Fuel Capital, Lowercase Capital, Redpoint Ventures, Reed Elsevier, Salesforce.com, and SV Angel. Individual investors like Derek Collison, John Dahl, Erik Frenkiel, and Jeremy LaTrasse are also on board.

Which drives home the fact that Heavybit isn’t a charity—the idea is to profit by having a hand in the evolution of the whole developer-tools industry, and making it as big as it can be. Given that there are so many corners of business operations and consumer life that have yet to benefit from cloud technology, the developer tools industry would seem to be all upside at the moment. The only real question is how big it can get.

On that point, you won’t be surprised to learn, Lindenbaum has a theory. On a graph of time versus profit margins, he believes, most stratified industries eventually reach a peak.

The more players there are at the bottom of the supply chain, the more everything gets commoditized, meaning that companies are forced to start competing on price rather than innovative features. This commoditization works its way up the chain until even the consumer-facing businesses at the top have run out of new things to do. Margins collapse, companies are forced to merge and consolidate, and industries end up back in a state of high vertical integration and low innovation. (Think of the cable and telephone companies, for instance.)

Ideally, then, if you’re building a new stratified industry like cloud software, you want to get to the peak of the curve as fast as you can, and then prolong your stay there. “This hill is a fact of life; the difference is that we think we can control how high the peak is, and how long it takes us to get there,” Lindenbaum says. “The key is finding new pockets of innovation who continue to press the peak up for everyone. I don’t think you can stay up there forever, but as long as people are willing to look for value and innovation and not worry about what the competitors are doing, that is how you maximize the value under the curve.”

Lindenbaum hopes that Heavybit, like Heroku before it, will offer a good observation post from which to guide the climb.

Wade Roush is a freelance science and technology journalist and the producer and host of the podcast Soonish. Follow @soonishpodcast

Trending on Xconomy