“Manufacturing hardware is hard!”, you’ll hear the common refrain from investors, entrepreneurs, and engineers. Products need to be designed for manufacturability, raw-materials and parts sourced. Work cells are designed for assembly and repair, and industrial operations are engineered to scale volumes without sacrificing quality. All of this has to be orchestrated with pitch perfection to deliver delightful products, on time, with low defect rates and in a capital efficient manner. Moreover, for all this risk that manufacturers and factory operators take, feedback loops of “is this line working well?” are painstakingly slow. It’s the opposite of ‘Agile’ or ‘Lean’ which are the dominant paradigms of software development and operations today.
A brief history of Agile
Software engineering wasn’t always this Agile; either for writing code or maintaining and operating it. At the turn of the 21st century, a group of developers waged a Lutheran insurrection against the “waterfall” model of software development, penning the Manifesto for Agile Software Development. Following in its footsteps, in 2013, Gene Kim published The Phoenix Project, formalizing the practice of DevOps to nimbly maintain and operate complex software systems and the infrastructure they run on. These operational practices are largely responsible for the proliferation of rapidly evolving and reliable software systems that we take for granted today.
This era also saw a class of companies – Splunk, Atlassian, Slack, Pagerduty – emerge to make software for agile software development and operations teams. Similar companies are yet to drive agility for manufacturing operations. Don’t get me wrong – certain parts of the manufacturing organization have indeed reaped the fruits of software-driven agility project lifecycle management (PLM) software has transformed product development; computer aided design (CAD) software has revolutionized design; enterprise resource planning (ERP) has made finance and accounting more efficient.
Toward an Agile manufacturing
However, the practices and software powering industrial manufacturing operations have largely remained unchanged since the 90s. Where present, operations are driven by multi-year inflexible and centrally-planned manufacturing execution systems (MES) transformation projects. These make sense only for the largest manufacturers, not for their parts suppliers, and they become obsolete nearly as soon as they are implemented. Visit any factory and you’ll see shop floor processes, from work instructions guiding complex assemblies to quality checks, done with pen and paper. Data is collected and analyzed manually.
All of this is particularly ironic when you realize that both the Agile and DevOps movements of developing and operating software borrow heavily from Kanban and The Goal – cornerstones of lean manufacturing ops and industrial engineering. The Agile processes are present on the shopfloor, yet absent from the software systems that power it.
This begs the question: What if instead of top-down planning, manufacturing workers and industrial engineers could work with small agile IT teams to test-out innovations to drive higher throughput, lower lead-times, improve yield, reduce waste and improve quality? All from the shop floor, one manufacturing line at a time. How much value could manufacturers unlock with an agile anti-MES?
Tulip powers Agile, bottom-up transformations
We met Natan Linder, Tulip’s CEO, early in the company’s life and at the time, struggled to realize how powerful the simple factory kit and work instruction software could be. Since reconnecting with Natan, Rony and the broader Tulip team last year, we have seen the power of bottoms-up adoption driven by shop-floor teams. Tulip’s work-instructions software has grown to integrate data from a plethora of tools and sensors, turning Tulip into a platform for any factory to build its own manufacturing apps.
Tulip’s manufacturing app platform arrives for manufacturing ops teams at a time when no-code platforms like Salesforce, Marketo, ServiceNow, and Zendesk have similarly transformed the roles of sales ops, marketing ops, IT ops, and customer-services teams. Tulip puts the shop-floor operations team, not central IT, at the forefront of industrial engineering decision making.
We saw the financial impact to manufacturers from the unprecedented operational visibility and flexibility between increasingly complex machines and tools, and the people who operate them. And we understood this as a cultural shift for manufacturing ops teams to drive agility at work. With Tulip, shop floor teams build, edit, and operate these manufacturing apps in the same way Splunk and Atlassian changed the game for software/IT ops teams. Vertex led the company’s Series B round in early 2019, joined by existing investors including NEA and Pitango, to power this revolution.
We could go on ad nauseum about all the value of IoT and AI that will be unlocked by this process data. But our heroes, the manufacturing ops teams, prefer that we eschew the buzzwords for real-world apps that actually drive operational metrics. Next week, Tulip will be on the AWS Pavilion at Hannover Messe, showcasing some of these real-world applications and unveiling capabilities that leverage cutting edge research in machine learning. Drop by, because it’s only Day 1 for agile manufacturing apps.
Sandeep Bhadra is a partner at Vertex Ventures and a Tulip board member. He holds a B.Tech from IIT Madras and a Ph.D. from the University of Texas at Austin in Electrical Engineering, and an MBA from INSEAD. You can follow him on Twitter and Linkedin.