September 7th, 2012

APIs for Manufacturing

Perhaps you too remember that crayon-factory segment from Sesame Street. The one where they show dramatic shots of hot orange wax getting poured into vats and molds and eventually labeled and boxed up into neat packages of the rainbow, while Skrillex-junior music played in the background. This was my favorite segment on Sesame Street, evidenced mostly by the fact that it’s really the only one I remember. Not to worry if you have no recollection, I found it for you on YouTube.

Some people freak out when they see movie sets, or space stations, or fire trucks, but for me it has always been factories. I spent some time 2005-2008 wandering in and out of factories in Shenzhen, Dongguan and Chaozhou and what was most surprising was how much is still made by hand. It is not all robots, even remotely.

And modern supply chains were built around that very premise – that hands were needed, and the best way to compete and scale was to find the cheaper hands. But some very interesting rumblings on the periphery of mainstream production (specifically in electronics) indicates not that the need for hands is going away, but that the entire system is evolving in a much more fundamental way.

The latest buzz-phrase-of-late seems to be “hardware is the new software” – and from what I can tell, the significance rests in the idea that the formerly painful process of making software has become a mostly democratized affair that has created many billion dollar companies along the way. The same will soon happen in hardware. And “soon” is used very loosely here. Could be 2, 12, 20, years down the road (ok probably not 2), but it seems near-inevitable that the shift will create many new and interesting companies along the way.

I titled this post “APIs for Manufaucturing” because that’s what I hope will happen; in many ways the changes are already underway. The dream is that manufacturing will eventually be democratized through a series of APIs. They will make it easy for aspirational producers to access expensive production devices that they would not be able to get access to on their own. Because sadly manufacturing is still extremely difficult and out of reach for all but the most tenacious producers. Open-source hardware and consumer-friendly 3D printing is slowly changing that, but getting any sort of scale remains quite difficult.

A great read from yesterday is Ben Kaufman’s blog post on Quirky’s recent $68m fundraise. Perhaps the most insightful observation for me was from this reaction piece: “After several trips to China, Kaufman realized that he actually needed to master a list of “50 or 60″ disciplines to make his product a reality”. It is these 50 or 60 disciplines that will one by one become companies in the still-nascent MaaS market (manufacturing-as-a-service, a term not-yet widespread but still used). Perhaps that’s why this is such an exciting time in hardware – the changes are materializing but are nowhere close to realized.

Mass market to MaaS market? Fuck yeah. The fun is just beginning.

Share This Post:


  • Have you seen DARPA's Adaptive Vehicle Make (AVM) project?
    They want to create an open source system of systems that would allow remote digital collaboration so that a vehicle (or anything really) could be designed in simulation and then the parts farmed out to various manufacturing facilities in real time.

  • One challenge that "APIs for Manufacturing" will hopefully solve is in easier interface design for hardware design tools. While design tools have come leaps and bounds, and super-simple proprietary tools have been around since the late 1990s (Express PCB, eMachineShop), giving developers APIs opens things up multiple fold. At a minimum, it opens the doors for UX hotshots to take a stab at designing CAD tools that more people can understand and use, and offer them at an more affordable price to the aspiring producer.

    This is particularly exciting given the multi-modal-input devices we now have available to us off the shelf (ie: tablets, touch, 3D).

    Granted, most people versed in hardware design will be versed with a PCB layout tool, HDL, etc, and most people versed in mechanical design will be versed in a mechanical CAD package, but if atoms are truly going to become the new bits, then the tools need to get easier and cheaper too.

    For aspiring producers, the daunting climb of a months-long learning curve and thousands of dollars licensing fees for say SolidWorks or Mentor Graphics is one of barriers that the APIs will hopefully tear down.

  • Would love to hear your thoughts on how services like Shapeways fit into this new world of MaaS?

  • dkural

    Scaling in software is also very difficult. The initial infrastructure for facebook / twitter that validated the market bears no resemblance to the scaled-up version.  Scaling is only necessary for mass manufacturing - but one can still validate an idea / make small batches of infinitely diverse range of products without it.  This is no different than software. 

  • Ben Garvey

    Despite advances in robotics, a human being can still be incredibly productive in manufacturing!  The biggest advantage of humans vs machines is that humans are nearly endlessly flexible in what they can do.  Clever ones can even figure out ways to do their jobs better.  

    Machines can perform incredibly fast, so manufacturing of goods in high demand can be mechanized using a variety of machines.  These machines can be changed over for different products, but for the most part are designed with one or two operations in mind.  In the computer analogy, these are like the big code breaking machines from WW2 before they came up with programmable ones.

    Robots are another step above that because they be reprogrammed to do different things, but once they are in place, if you want to make anything drastically different you have to go there and move the robots around and reconfigure them.  

    To get anywhere near the flexibility we have with software, robots and machines will have to get much more flexible.  There's no Moore's law in manufacturing, but continuous improvement will get us there eventually.  One point where the analogy fails is distance.  I can write a web app that uses SaaS companies like Stripe (credit cards) and Postmark (transactional email) and I don't have to worry about where they're located.  In a MaaS where there are transportation costs, it matters where my parts and materials come from.  That's why companies  buy these "expensive production devices."

    I work in manufacturing (  We help companies make their automated packaging systems more efficient.

blog comments powered by Disqus