Systems of Intelligence
Definition and Thesis
Over the past 36 months, it’s been clear that opportunities innovating around manufacturing have percolated to the top. Here we explain why we’re excited about the future of Industrial Technologies, how we look at this world, what we look for in Founders building the next generation of solutions, and how to build a system of intelligence in a legacy industry.
“Industry 4.0 is the current trend of automation and data exchange in manufacturing technologies including Cyber-Physical Systems, IoT, and Cloud Computing.” Industry 4.0 is made possible through the convergence of several trends: 1) seamless communication between machines, systems and humans, 2) information saving, sharing, and transparency, 3) technology supporting humans (Co-Bots), and 4) decentralization of systems and processes (i.e. On Demand Manufacturing). A key feature enabling the growth of Industry 4.0 is the capture and useful deployment of data in providing meaningful and actionable insights, in real time. This is the next logical iteration of the manufacturing process following Industry 3.0 in the late 1960’s and 70’s, which was focused on Computer and Automation.
When separate functions across manufacturing build a system of intelligence together, factories become more responsive, more efficient, safer, and forwardly defensible. If the factories of today are akin to the data centers of yesterday, then they must make valuable sense of the new big data available. They must link systems together and produce timely, coherent insights that lead to defensibility.
“Said another way, Industry 3.0 was about things acting on their own; 4.0 is about things acting together.
On-Demand Manufacturing Takes Center Stage
On-demand manufacturing (e.g. Cloud or Digital Manufacturing) encompasses a variety of companies, concepts, and competencies, laying the foundation for a system of intelligence. At its core, Digital Manufacturing allows for small batch and customized manufacturing, and companies wanting to do “mass customization” quickly and affordably. This also includes rapid prototyping and part replacement manufacturing at the edge. It reduces the need for large minimum orders, decreases lead-time, and shrinks the long and expensive processes of shipping and inventory holding, which reduces the need for major CAPEX, thereby creating barriers to entry. If implemented cost effectively, incumbents no longer can hide behind their moat of massive production facilities to sell units at marginal price = marginal cost. Now, manufacturers can specialize in many things and outsource (either internally or externally) to other manufacturers while maintaining profit and service levels. The ability to connect to multiple on-demand manufacturers for support and specialization allows companies to be more responsive and builds a connected and dynamic system through optimized communication and workflows which establishes a virtuous cycle and thus creates a moat.
Within on-demand manufacturing, companies must focus on vertical or horizontal specialization. By specializing within a vertical, a company gains expertise and a reputation while being able to concentrate on one set of customers. An example is MacroFab (who has some of the best articles available), which produces only circuit boards, but handle everything from customer management, taking the quote, printing, shipping, etc., involved with an order.
Horizontal Generalization / Marketplace
Some companies (Hubs) connect the customer with a 3D printing service bureau that can best suit their needs and is local, reducing shipping costs and lead-time. Some bureaus assist with design by connecting customers to skilled CAD designers, similarly to 99designs offerings with digital and print graphics. Note we are excluding B2C companies, which exclude firms like Shapeways, iMaterialise, Sculpteo, and Ponoko, given the increased security and quality assurance expectations that Industrial customers have. Ultimately, these companies rely on scale to achieve breakout success – expect them to build out geographic hubs and differentiate on value-added customer service.
Value Chain Supporters
It is important to consider the full value/supply chain when assessing Digital Manufacturing (DM) opportunities, as there are many different processes involved to fulfill any manufacturing order:
- Customer education, marketing, and funneling to sale
- Placing an order (quotes and prototyping)
- Customer management (tracking, order, and project management)
- Sourcing materials, determining specs, and locating manufacturer (if not in-house)
- QA including warranty & insurance
- Actual manufacturing/printing
- Inventory storage and management
- Shipping and fulfillment; logistics
- Invoicing and payments
- Repairs and support; upgrades and replacement parts
Capitalizing on data insights from the entire value chain is referred to as the Digital Thread, and unlocks a variety of opportunities, such as enabling the DM process from design to print head (Identify3D).
As mentioned above, many companies do all of these activities or unify the value chain, others just print, and some are simple marketplaces. Many of the activities above are completed with traditional tools such as Salesforce, but opportunities exist for startups, particularly in functions 4, 5, 7 and 8 above. Part of the allure of DM is locality and speed, which creates synergies with last mile companies such as AxleHire.
The Rise of Collaborative Robots (Co-Bots)
Robots have long been part of the manufacturing process. Think of the automobile or airplane assembly line with their rows of giant robotic arms moving and assembling parts, and programming dashboards. However, these massive, multi-million dollar robots must operate in their own space, unsafe for direct collaboration with humans. They cannot see or sense when they put a human in harm's way, and are generally not easy to re-program for a different task.
But this perception doesn’t necessarily match reality, as up to 90% of manufacturing is at a much smaller scale, where a small robot working WITH a human can increase productivity and safety in the workplace. Enter collaborative, or co-working robots (co-bots), like Carbon Robotics.
Co-bots are still only working with humans in a limited capacity. Although more collaborative than traditional robots that are separated from humans, co-bots are still working on their own more than they are with humans. Due to safety concerns, co-bots are limited in how big they can be and thus, how much weight they can lift, so the applications are still limited in scope. Additionally, customers must consider that while robotic arms have an extensive range and can handle heavy payloads, human hands are still more dexterous and provide functionality and detail not yet available in co-bots.
The most important factor for the rise and proliferation of co-bots is safety, of which there are two main approaches. First, is a complete shutdown when a person gets too close or touches the bot, which requires a full line restart once the co-bot is cleared to proceed. This approach is generally applied in heavier scale applications, and thus is incremental relative to the innovations we’re seeing within robotics. Another approach is for co-bots to slow down when a person approaches and stop when they get too close, with the ability to quickly restart. With promising stats showing that the cooperative process reduces human idle time by 85%, we believe co-bots will continue to move towards the slow-down method, allowing for increased collaboration and the reduction of shut-down time.
Co-Bot as a Platform
Co-bots are designed to be safe around people, can be trained for a variety of tasks (easy to program), can be moved around and learn different tasks, are intended to augment instead of replace, and are cheaper and simpler than traditional robots. With the rise of machine learning and IoT, co-bots will continue to learn and be easier to program. This will lead to co-bot hardware becoming commoditized and create opportunities for API plugins that reduce specialization and the need for training, while adding innumerable capabilities. Co-bots have the potential to become an all-in-one tool for the factory floor. This tight collaboration between humans, co-bots, and smart data collection (with IoT) can create a valuable and deep system of intelligence.
IoT and Leveraging Data
The rise of IoT exemplifies the promise of Big Data actually paying dividends. Manufacturing is an industry that operates on relatively low margins, has a relatively repeatable process, and firms are constantly looking for any incremental improvement to cut costs, reduce error, or speed up throughput. Previously these changes were made through observation, anecdotal evidence, and gut feeling.
The term and concepts of IoT date back to the 1990s. Though IoT investment has been climbing steadily, it has exploded over the last few years and will increase from around $40b in 2017 to almost $80b in 2020. The ability to create insights about manufacturing starts with the ability to collect data. IoT chips are the first step in collecting that data. Currently, these are mostly sensors that track things such as temperature, heat, vibration, etc. but could evolve into something more akin to a computer chip.
Industrial Analytics (Prescriptive)
Industrial Analytics (IA) refers to gathering and utilizing data across the entire product lifecycle and, according to Forbes, organizations believe IA is responsible for “increased revenue (33.1%), increased customer satisfaction (22.1%) and increased product quality (11%)”. Platforms that can digitize and institutionalize data can then discover patterns such as bottlenecks, where errors are occurring and why, faulty machines, and more. These tools can provide insights into resource use, count errors, and show variation data, which helps with quarterly and annual planning.
Machine Learning (Predictive)
One of the great things about data for manufacturing is that machines are constantly collecting data and the data should will be consistent. Compared to something like Autonomous Driving, where the car is not on the same spot on the same road all the time, machines in a factory are performing the same task in the same environment constantly. That means the data are easily classifiable, numerous, and consistent. These are conditions where machine learning is most effective, so manufacturing, aided by IoT, is the perfect use-case. Machine Learning algorithms can proactively shut down machines before they break, help plan the most efficient flow, and conserve power by turning off machines that don’t need to be on.
“When IoT enabled devices are paired with industrial analytics tools and applied machine learning, manufacturers can better utilize data for decision making and forecasting, creating a system of intelligence for the factory floor.
Positive Traits to Consider
The Share Economy
At Bee Partners, we see a pattern emerging in Industry 4.0 that is borrowed from the share economy - renting is the way of the future. Renting, service agreements, API and SaaS type models are replacing the traditional buy and depreciate method of industrial manufacturing and is opening up opportunities for small businesses and startups that previously were excluded due to lack of scale. This leads to changing business models and potential for technology to revolutionize the entire value chain. Robots and AI will continue to insert themselves seamlessly into the processes of construction, light and heavy scale manufacturing, logistics and beyond.
Any successful startup starts with a beachhead or wedge, and expands its offering from there. Any good founder must have considered the next move (or moves, optionality is fantastic) from there, and be laying the foundation now. For example, one must either focus on a particular function (marketing or analytics) or a particular industry (appliances) and then expand offerings from there.
Without question and given the rise of the Industrial IOT, security must be at the foundation of any new industrial offering. Indirect attacks have become commonplace, so we will always consider the security element of new opportunities when paired with Industry 4.0.
Empathy for the Worker
While this may seem anachronistic, the rise in technology has only magnified the erosion of middle class employment opportunities. New entrants must have a sales conscience, and often be willing to retrain workers, raising their skillset. Skycatch’s retraining of surveyors seeking ground truth highlights that job re-creation must add additional value and cannot be superficial. Furthermore, companies should attempt to predict and leverage how new industries and technologies will create new jobs that do not exist today.
Plus + 1
As we’ve seen in other industries with mature stacks, simply adding another layer into the workflow, especially a mature one, is rarely sufficient. These mature industries don’t want another 3rd party sensor or another login. Truly innovative solutions must have something additional to offer, such as eliminating the need to move parts around (logistics) or including the materials into the SaaS offering. It is also important to not ask people to pay for a new service, unless it provides significant incremental value.
Pitfalls to Avoid
Solutions Above the Natural Skillset
We are still making things. We still live in a world where humans must control and interact with the robots, and enjoy the outputs of our increased productivity. Technology based solutions that seem to ignore this, or build to a co-worker and end user that cannot make the leap up, will struggle with customer adoption and attracting down-stream capital.
Ignoring the Current Value
Whether as a result of ignorance surrounding the value chain or the desire to be an outsider disrupting, Founders cannot dismiss the fact that the current value chain is mature, with deeply ingrained co-dependencies and relationships. We believe that it will be Experienced Founders with deep market insights who will capture the value in much of the manufacturing element of Industry 4.0, not necessarily a technologist looking to replace the existing players. At Bee Partners we weight heavily the notion of Founder-Business fit.
Displacement of Jobs for Subpar Work
The eroding middle class is already suspicious that they will be replaced by automation. Any new entrant that results in subpar work will never find a foothold. As investors, we must talk to the potential early adopters to understand if the technology solution proposed will truly be better, faster, and cheaper. [Note: this displacement will also apply to entry level jobs in almost every industry, including retail as well as higher order roles within banking and law.]
Lack of Transition Facilitation
There exists much inertia on the shop floor. Factories and the personnel therein have been operating in a specific protocol for years, even as new and better technologies enter the workflow. Any piecework solution that does not consider getting from current to a future steady state will face adoption headwinds.
Facilitating transition will be paramount as Silicon Valley further inserts its culture into the future of manufacturing. The disruption of the current human interface will continue, which will redefine many aspects of the value chain and provide opportunities to build moats behind forward looking systems of intelligence. When set in motion, this transformation happens rapidly and will require justification to the existing stakeholders.
We expect this accelerated transformation to amplify calls for saving manufacturing jobs, to focus on making them safer, and to highlight how these emerging systems and technologies can make on-shoring profitable. Further, new themes will emerge (redefining logistics, inserting narrow AI, or utilizing computer vision) that will draw talented Founders and investors to dedicate their livelihoods to affect positive outcomes. However, we do expect humans to stay in the loop for the foreseeable future, and for overall economic benefit to continue driving the underlying transformation. At Bee Partners, will continue to explore and invest in these frontier curve opportunities surrounding Industry 4.0. We have always backed Founders executing on their deep market insights, now with a keen eye towards those building lasting systems of intelligence.
Associate, Bee Partners
Partner, Bee Partners