Where Industry Meets Innovation

  • Contact Us
  • sign in Sign In
  • Sign in with certificate
mit campus


Search News

  • View All
  • ILP News
  • MIT Research News
  • MIT Sloan Managemet Review
  • Technology Review
  • Startup Exchange


June 22, 2017


31 Results | Page 1 | 2 | Last | Next
June 1, 2017

Predictive Analytics for Retailers

Celect is helping retailers reveal true demand, leverage existing data, and optimize their stock using predictive analytics and machine learning technology.

John Andrews

“Retail is clearly going through a transformation right now,” says John Andrews, CEO of Celect, a predictive analytics machine learning technology company focused on retail. “The reality is that stores are not going away. Customers are looking for experiences, and part of that experience is to touch the product and to try the product on,” he says. Improving the overall customer experience means understanding at a more precise and granular level how customers choose between an assortment of products. The Celect Choice Engine helps retailers identify and predict what a customer faced with an assortment of products is likely to buy.

Customer Choice Modeling
“The context of every customer’s decision delivers a very rich signal in terms of how individual customers choose,” says Andrews. At the core of this is customer choice modeling. When a customer walks into a store and buys a product, retailers use transaction-level information to personalize the experience and to optimize operations within their store. “But what if in addition to the customer’s purchase, you also knew what their options were?” asks Andrews. Said another way: What if you knew what customers didn’t buy?

That’s where the predictive piece comes in. One of the key challenges retailers have is around the problem of sparse data. In terms of identifying buying patterns of an individual customer over the course of a year, a retailer may only have one or two data points. Gathering data on a customer’s buying patterns, their browse history, and what they put in or remove from their shopping cart produces a very robust model that allows Celect to predict the likelihood of future behavior. “Based on this,” says Andrews, “what are the right products to put in front of an individual customer and in what quantity?”

Inventory Portfolio Optimization Challenge
At the highest level, the inventory portfolio optimization challenge is the problem Celect is solving: How much of an individual product does a retailer need to buy and how is that product going to interact with another product? “The complexity of that model can be baffling,” says Andrews. Until Celect came along, there was no real system that could handle such complexity. “We have a solution that can help merchants and planners identify what products they should be bringing into their assortment, and where they should be putting those particular products,” says Andrews. One of the core capabilities of the inventory portfolio is being able to identify how well a product will sell in the future based on history and purchases.

Inventory Portfolio Challenge
Around the idea of Inventory Portfolio Optimization is Celect’s longer-term vision. Be it online, direct, wholesale, or retail in-store experiences, Andrews sees a much bigger opportunity across the entire supply chain, from brand and manufacturers to distribution and retail. “At each step across that supply chain, we identify how much a retailer should be buying, at which distribution centers or fulfillment centers the product should be brought, how much of that product should go to each individual store, and in what assortment,” says Andrews.

Plan. Buy. Allocate. Fulfill.
Under the umbrella of Inventory Portfolio Optimization, Celect is focused on four core solutions, mapping directly to the process most retailers live by. The first, Plan Optimization, used in strategic planning and in merchandise financial planning, helps retailers identify how much they should be spending on specific departments, brands or styles, the demand for those products within their customer base, and in which stores they can sell those products.
The next module, Buy Optimization, helps retailers determine the demand for a product, whether they should be going big or buying small. “Getting that right early in the decision process is incredibly important in terms of what final revenue and markdown numbers are going to look like for a retailer at the end of the season,” says Andrews.

The third piece of optimizing inventories is Allocation Optimization. “Now, I’ve got an assortment of products. I know how much of each product I have. Where should I be allocating those products? Get the product into each store, in the right assortment and in the right number based on the buying patterns of customers in those stores,” says Andrews.

The final piece is the Order Fulfillment Optimization solution – the process of intelligently leveraging store inventories to fulfill online orders. Here, the retailer is trying to push as much of the inventory into their stores, use their stores as fulfillment centers, and then intelligently identify from which stores they should be shipping that product. Andrews says understanding the demand for a product over a course of a season can make for a much smarter decision in terms of which store to ship products from.

Using Your Data for Better Decisions
“Omni-channel—it’s probably an overused term—but it’s a real issue and challenge for retailers to figure out how to leverage every interaction point, every channel with a customer, and then be able to optimize across all of those different channels to provide the best experience to customers,” says Andrews. Whether customers want to buy something online and return it to a store, buy something online and pick it up in a store, or buy something in a store and mail it back, involves an enormous amount of complexity from an operational perspective. “At the end of the day,” he says, “it all comes down to getting the right product in front of the right customer at the right time.” As part of this transformation, retail is going evolve and change. Some retailers will have fewer stores; others will open more stores.

Andrews says you don’t need to be Amazon to use your data and make smarter decisions. “Predictive analytics and leveraging machine learning to supplement the decision-making is at the top of every retail executive’s priority list. They quite simply want to understand how to use it, how it gets integrated within their environment,” he says. As Celect has grown, so have the data points on how different retailers use information and science to help supplement decision-making, to help retailers make better decisions, to increase revenue, reduce stock-outs, and reduce markdowns within the customer experience. “The retailers who are able to truly understand how their customers are interacting with products and how the products are interacting with each other, and are then able to optimize on that are the ones who are going to win.”

About STEX25 and MIT’s Industrial Liaison Program (ILP)
STEX25 is a startup accelerator focused on fostering collaboration between MIT-connected startups and member companies of MIT’s Industrial Liaison Program (ILP). STEX25 is managed by MIT Startup Exchange, and its parent, the ILP. The ILP is a key player in making industrial connections for MIT, with over 220 of the world’s leading companies using their ILP memberships to advance research agendas at MIT.

May 30, 2017

MIT Spinoff Company Develops Self-Driving Technologies to Enable Safe, Sustainable, and Equitable Mobility Solutions

Sertac Karaman and Ramiro Almeida, co-founders of Optimus Ride, are developing self-driving technologies to enable safe, sustainable, and equitable mobility solutions.

Sertac Karaman
Optimus Ride

As we strive towards an inevitable future where autonomous vehicles dominate the urban landscape, transporting both people and goods with greater safety and efficiency, Optimus Ride is well positioned to play a significant role in what must be a continual evolution. As of October 2016, they have completed a series seed investment of $5.25M, co-led by NextView Ventures and FirstMark Capital. Other key investors in this latest round of funding that will allow Optimus Ride to accelerate the development of its autonomous vehicle technology systems include NVIDIA GPU Ventures, Nicco Mele, the director of the Shorenstein Center at Harvard University, and Joi Ito, the director of the MIT Media Lab. But what sets Optimus Ride apart from other would-be innovators in what is rapidly becoming a congested marketplace?

President and chief scientist, Sertac Karaman points first and foremost to the founding team and their supporting players. “Optimus Ride is a true MIT spinoff. The whole team came together at MIT, representing multiple departments including the Media Lab, CSAIL, Aeronautics, and Sloan,” says Karaman, who is himself an associate professor at MIT in the Aeronautics and Astronautics department, as well as being affiliated with the Laboratory for Information and Decision Systems and the Institute for Data Systems and Society. Karaman and his co-founders Ramiro Almeida, Ryan Chin, Albert Huang, and Jenny Larios Berlin, are a formidable team. Together they boast over 30 years of interdisciplinary university research in self-driving technologies, electric vehicles, and Mobility-on-Demand Systems, not to mention a decade of industrial and entrepreneurial experience that combines manufacturing robots, urban design, and shared vehicle fleet management.

“We are on the edge of a transportation revolution that will be enabled, in part, by technology and robotics,” says Karaman. He continues, “I believe that Optimus Ride is very well-positioned to be one of the most important players in this domain, as we build self-driving vehicle technologies that will create new transportation systems that will truly transform the industry and have a global impact commensurate with the breakthrough of trains, the affordable car, or the airplane.” As one would expect, with MIT-based experts from a diverse set of disciplines at the helm, Optimus Ride leverages the latest advances in complex sensor fusion, computer vision, and machine learning to develop its systems. Karaman makes it very clear that the technology is just one aspect of a larger whole. The Optimus Ride vision is to provide safe, sustainable, and equitable mobility solutions. These are of course multidimensional terms that take into account a variety of factors including energy efficiency, societal constraints, affordability, the ethical implications of writing code for autonomous vehicles, and even aesthetics, all of which Karaman and the team at Optimus Ride consider in an effort to make transportation and new transportation systems more enjoyable for everyone.

They have worked on a range of different autonomous vehicles, from golf carts to fork lifts. The end goal remains the same: getting the technology to the end user where it can be utilized in urban environments and beyond. With this in mind, they have just moved into a 20,000 sq. ft. facility in the Boston Seaport District that allows them to efficiently design, build, test, and develop their systems further. And they are looking to grow. New, though as yet unnamed pilot locations are in the works. Co-founder Ramiro Almeida says, “As we consider the finer details, we may realize that a system that serves a certain society well may not serve another as well. At Optimus Ride we pay attention to this, and we will be deploying a number of pilots in different places to be able to better identify and understand the variables that make a big difference in different locations as we develop and deploy our technology in these domains.” He continues, “Our technology has the potential to improve quality of life for millions of people around the world. Transportation systems, like taxis, buses, and trains have been around for decades, and we have experienced minor improvements throughout the years. Robotics technologies present a major opportunity to design new systems that consider data and user needs to provide the most efficient solutions that can be adapted at a relatively low cost to any urban environment.”

Karaman recognizes there are many different challenges facing the industry. From developing scalable business models to urban architecture, as well as policy and law. As a technologist he admits he is prone to wanting things to move quickly. That said, looking back on his decade of work in the industry, he is pleased with how rapidly things have progressed, especially in the technology domain. “The kind of technology that we are using is really diverse,” he says. “It’s not just a particular algorithm that enables everything, but it is so many. There are software implementations that are very complex, and even the computers they run on, and the sensors that enable this—they are all coming together at a very fast pace.” And despite the challenges, he thinks we will be surprised at just how quickly the technology will become available. Though he posits, it might not be in the way we expect.

As Optimus Ride reshapes the future, Karaman reflects on what he refers to as “a deep relationship between academia, industry, and innovation.” He recalls working on the DARPA Urban challenge. The public looked at their driverless car as little more than a novel academic project developed in an echo chamber. “Fast forward ten years” he says, “and people recognize that self-driving vehicles are a part of our future, and an essential technology we are going to rely on for a number of transportation and logistics needs.” As MIT has developed a reputation for producing successful startups and the ecosystem has grown, he points to ILP initiatives like STEX as invaluable tools for strengthening the MIT startup community—not only connecting new tech-based ventures with one another, but with industry and investors capable of providing opportunities for start-ups that began as research projects and demonstrations to become marketable products, one of which might just become the next big innovation to positively transform our lives.

About STEX25 and MIT’s Industrial Liaison Program (ILP)
STEX25 is a startup accelerator focused on fostering collaboration between MIT-connected startups and member companies of MIT’s Industrial Liaison Program (ILP). STEX25 is managed by MIT Startup Exchange, and its parent, the ILP. The ILP is a key player in making industrial connections for MIT, with over 220 of the world’s leading companies using their ILP memberships to advance research agendas at MIT.

April 24, 2017

Winning Back Unresponsive Customers with Artificial Intelligence and Machine Learning

Alan Ringvald, CEO and Co-Founder of Relativity6, utilizes proprietary behavioral listening algorithms to predict when and what a company’s most profitable lapsed customers will repurchase with a greater than 80 percent accuracy rate.

Alan Ringvald
Cofounder and CEO

Though Relativity6 was launched less than a year ago, CEO Alan Ringvald attributes much of their early success to the fact that for he and co-founder and CTO Abraham Rodriguez, decoding unresponsive customers’ behavioral economics is a longtime obsession. While both had prior success separately in the start-up realm, Ringvald as a co-founder of Superdigital and Rodriguez as a co-founder of Northrend Centinela, the roots of their current venture Relativity6, dates back to their time as students at MIT where they co-wrote their master’s thesis on reactivating unresponsive customers through machine learning. While in school they were also the recipients of the MIT Sandbox Innovation Fund, which not only provided them with their first funding, but also allowed them to pitch their product for the first time, and was, as Ringvald puts it, the springboard for what would eventually become Relativity6.

Relativity6 utilizes proprietary behavioral listening algorithms to predict when and what a company’s most profitable lapsed customers will repurchase with a greater than 80 percent accuracy rate. “The technology behind Relativity6 is quite streamlined at this point,” he says. “Initially we ask for purely behavioral data from a company. We don’t want names, we don’t want emails, nothing personal, we just need a unique customer identifier because what we are doing is analyzing internal past purchasing behaviors.” This focus on behavioral data means that stringent privacy policies have no bearing on what Relativity6 does. In fact, Ringvald is quick to point out that they take privacy very seriously, and there are several cybersecurity PhD’s on staff. Not only do they not want or need personal information such as names, email addresses or credit card numbers, but that type of information doesn’t help their model. “All we need is a unique customer identifier and we are good to go,” he says. The raw data is then run through their machine learning algorithms, and what emerges are predictions of which lapsed customers will repurchase and which product or service they are most likely to repurchase. In addition, the process allows them to predict through which channel they are most likely to reengage, be it email, phone, or catalogue. The client is provided with these predictions and then uses them to reengage former customers. The truly elegant aspect of the model as designed by Rodriguez is that, as Ringvald puts it, “Whether we are right or wrong, the model retrains itself; it is the beauty of machine learning. It learns whether it was accurate or not and is able to retrain and be more accurate next time around. And that is the process that repeats itself until we reach the 80 percent accuracy rate.”

In terms of customer base, companies that have participated in Relativity6 pilots vary greatly, from those that have only 1,000 customers to those with upwards of 50 million. Ringvald stresses that that company size and customer base are not key predictors for what Relativity6 does so successfully. Rather, all he and his team of MIT professors, data scientists, behavioral economists, and business strategists need is 18 months of back-data for their algorithms to understand past behaviors and thereby predict future behaviors. The numbers: Relativity6 finds out who will repurchase with a 95 percent model accuracy rate; when they will repurchase with an 80 percent match rate between predicted and actual lapsed customer future purposes; and what they will repurchase with 2-5 percent monthly conversion rates of total lapsed customers.

Thus far, Relativity6 has worked with companies of various sizes from a wide a range of industries. For example, Nutraclick, a technology driven company that provides leading health and wellness products, engaged Relativity6 to reactivate customers from their subscription service, and tripled their ROI in just one month. Other case studies done with companies including Zipcar, Coachup and Magellan Jets have yielded similarly positive results. “In terms of an ideal customer,” says Ringvald, “Relativity6 can help any organization that has been around for more than two to three years, has customers that haven’t purchased in a long period of time, and has behavioral purchase data. Literally any organization that has kept their data and has enough customers for us to be able to analyze.” This includes financial institutions, insurance agencies, hospitals, and retailers, but extends to political organizations, universities and nonprofits in terms of gifting and donations.

And with a seemingly endless list of potential clients looking to benefit from working with Relativity6 and their machine learning algorithms, the future looks very bright for Alan Ringvald and his team. They recently joined AI world leaders NVIDIA’s AI Inception program, and have even partnered with them on an external basis. Ringvald is also excited that Relativity6 has joined the ranks of STEX25, and cites the partnership with MIT ILP as particularly fruitful. “ILP has been an instrumental part of Relativity6,” he says. “We have gotten incredible support from the staff, and have already started working with several companies in the network.” And people are paying attention. Ringvald, on behalf of Relativity6, presented at the MIT Consumer Dynamics Conference (January 2017). Most recently, they were tapped to present at the MIT Silicon Valley Showcase (February 2017) hosted by Google. Relativity6 is picking up a significant head of steam. The accuracy of their algorithms and the success rates of their clients are proof of their achievements thus far. Ringvald says the next frontier is being able to automate the process of predicting why a customer defected in the first place. And Relativity6 is already hard at work on this next piece of the puzzle.

About STEX25 and MIT’s Industrial Liaison Program (ILP)
STEX25 is a startup accelerator focused on fostering collaboration between MIT-connected startups and member companies of MIT’s Industrial Liaison Program (ILP). STEX25 is managed by MIT Startup Exchange, and its parent, the ILP. The ILP is a key player in making industrial connections for MIT, with over 220 of the world’s leading companies using their ILP memberships to advance research agendas at MIT.

March 28, 2017

Small Satellites with Huge Capabilities Propelled by Ion Engines

Accion provides efficient alternative to rocket propulsion
Natalya Baily
Accion CEO and co-founder of Accion Systems Natalya Bailey (formerly Brikner) was interested in math and aliens in high school. After majoring in Aerospace engineering, she interned with “a big Aerospace giant” where she says, “hardly anything new ever went into or left that building.” So, she went to grad school, finished her Master’s degree in nine months and at the behest of her adviser, who gave her his credit card, told her to book eight trips to MIT and go sit outside the lab of two professors (one of whom would become her PhD adviser), and landed at the Space Propulsion lab. There, along with Accion’s co-founder Louis Perna, Bailey discovered that the ion source they were working on was really great at propelling small satellites in space. “It just so happened in the past decade the industry was trending toward smaller and smaller satellites that were lacking in any sort of efficient satellite propulsion technology.” It caught the attention of industry, and in 2014, while they were still students, Accion was formed.

How it works
Ion engines have been around since the 1950s and 60s, and like any sort of rocket propulsion technology, they propel forward by expelling matter out of the back of a spacecraft. Rather than relying on chemical propulsion typically used by the big rockets of Elon Musk, Accion’s engine propulsion technology, relies on electricity. “We accelerate some matter—it happens to be ions—to high speeds and push those out the back of the spacecraft. If you picture an astronaut riding on the back of a satellite throwing tennis balls off the back of it, each time he or she throws a tennis ball off the back it pushes the satellite in the opposite direction. That’s effectively what we do with ions,” says Bailey.

Who Uses Them
“The world relies on space more than most people realize,” she adds. Industries and applications from Direct TV to Sirius radio to GPS—even Monsanto—use satellites. “Accion’s technology is geared toward these smaller satellites that are starting to come online thanks in large part to Moore’s Law, which is making it possible to build very small capable satellites,” says Bailey. Previously, only whole countries or governments could afford to launch satellites, but now that they’re becoming smaller and cheaper to manufacture and launch, they are having the effect of making space more accessible and affordable to industry. New industries, like big financial institutions, are using satellite imagery to predict futures prices. Satellite constellations are going into orbit monitoring ships and tracking other assets as they move across the world. Another application Accion finds exciting is a mobile breast cancer clinic that used to wait ninety days before it would take the data and upload it at a doctor’s office. Since signing with a satellite services provider, the clinic can now upload the data in real time. She says, “For some patients that makes all the difference.”

The Product
For now, Bailey and her team are trying to focus on one main product, which she says, “was designed using the most boring manufacturing techniques we could think of,” and do it really well. That product uses three main components: power electronics, a propellant supply system, and a thruster head. The power electronics are sent out to a traditional PCB (Printed Circuit Board) house, returned to and tested by Accion. The propellant supply system, which Bailey describes as similar to a Tupperware container, allows Accion to forego the use of any big pressurized tanks, pumps or valves. “We get to store all of our propellant for any mission in this plastic box which really simplifies our design and any sort of reliability concerns over a lifetime.” The thruster head is made using microelectromechanical systems techniques—the same processes used in the computer processor industry. “These we send out and get back in batches of around two-hundred—our numbers are going up there. All of these processes are very mature. I think that’s going to be a key differentiator for us as far as cost down the road.”

In the very early days, Bailey says, the most challenging aspect of trying to start a space hardware company was the paradox of needing to build a working prototype before fundraising, getting money and selling things. “To build the prototype you need access to equipment and you need expensive materials and those things cost money. But to get the money you need the prototype; it’s just kind of this vicious cycle. We ended up trying to solve that problem and get to the next stage of Accion we have been doing kind of a hybrid venture-capital-government funding model.” She adds that Accion started with the assumption that a space start-up could sell only to other space start-ups. “When our product was first out the door, it lacked the flight heritage and all the reliability data that folks like Lockheed and Boeing and NASA would need to see. In reality we’ve actually found it to be the complete opposite. The bigger aerospace companies have internal R&D budgets that will evaporate if they don’t spend them. They’ve actually been our first customers and have been kind of eager to buy our very beta products.”

Accion is currently evaluating two paths in the long term. In the near term, it is a propulsion system provider—a component provider to a larger system. In the future and because it strongly believes it has something their customers genuinely need, Accion is considering a path to becoming a satellite services provider. In addition to making the propulsion system, it would also launch and operate the satellites, and provide communication services from satellites that already use its technology. Bailey says there are few times in an industry where there’s a schism in the space industry due to a perfect storm of technology and the political environment and various other factors. “It’s this wonderful opportunity for small companies and start-ups to come in and up-end the incumbents. Accion wants to move from our foot-in-the-door propulsion system model to eventually becoming a service provider.”

About STEX25 and MIT’s Industrial Liaison Program (ILP)
STEX25 is a startup accelerator focused on fostering collaboration between MIT-connected startups and member companies of MIT’s Industrial Liaison Program (ILP). STEX25 is managed by MIT Startup Exchange, and its parent, the ILP. The ILP is a key player in making industrial connections for MIT, with over 220 of the world’s leading companies using their ILP memberships to advance research agendas at MIT.
February 23, 2017

New Startup, Gamalon, Invents And Commercializes Revolutionary Machine Learning Technology

Ben Vigoda, the founder and CEO of a new startup, Gamalon, Inc., has created a technology that has automated the process of model writing, transforming the technology from a tool for data scientists into a full fledged machine learning technology that learns from data, by itself.

Ben Vigoda
Founder and CEO

Andrew Ng, one of the progenitors of Google Brain and an advocate for the resurgence of neural networks in recent years, told his origin story in a TED talk. He said that he tried to make a self-flying helicopter by programming a mathematical model of the helicopter, but no matter how hard he worked on his mathematical model of the aerodynamics and the helicopter mechanics, he could not get the helicopter to fly. Then, he decided to just rely on data from the helicopter’s sensors to train machine learning algorithms, and the helicopter flew amazingly well. Ng says that he learned from this experience to not try to write complicated models himself, and instead rely on deep learning algorithms.

Ben Vigoda, the founder and CEO of a new startup, Gamalon, Inc., in Cambridge, Mass. had almost the same formative experience as Andrew Ng, but he reached a completely different conclusion. At his previous MIT machine learning chip startup, Lyric Semiconductor, Inc., later acquired by Analog Devices, it took a long time and a lot of funds to develop working mathematical models for signal processing applications. But Vigoda decided not to give up on writing mathematical models. Instead, he decided that what was needed were tools for developing and debugging mathematical models and fitting them to data more quickly. Vigoda says, “Writing these kinds of models, fitting them to data, seeing what was wrong, and then improving the model was slow and arduous. It could take many weeks to try out one idea for improving a model, and often it would take years to get to a working system. For regular programming, we have compilers, debuggers, profilers, etc. all of this technology that makes it very easy to rapidly write and improve your programs that you are writing. This development environment enables agile programming. But for mathematical (Bayesian) models - the kinds that scientists write in order to fit their data, we didn’t have any analogous development tools.” Vigoda created Gamalon, and with funding from DARPA, set out to build these tools. “The first result was that we were able to build and test models much faster. We could write scientific (Bayesian) models and test them against the data almost instantly, and see which models and which parts of models help to explain the data most effectively.”

But that was just the beginning. The real ‘ah-ha’ moment came when they realized that they could begin to replace the human modeler. “We found that the development tools we had built for humans could actually be used to guide the computer to make its own changes to the models and autonomously test these new models. By automating the process of model writing, we transformed the technology from a tool for data scientists into a full fledged machine learning technology that learns from data, by itself. Already the system is performing very well on traditional machine learning tasks like image recognition and natural language processing.” Gamalon calls this new invention, Bayesian Program Synthesis (BPS), and it performs quite differently than deep learning. The company has a demonstration video where the system learns to recognize drawings in a side-by-side comparison with a drawing recognition app called “Quick, Draw!” from Google DeepMind. Gamalon’s BPS system learns from just a few training examples rather than thousands or millions, from one person rather than thousands, runs on an iPad rather than on hundreds of servers, and learns almost instantly rather than taking days, weeks or months to complete its learning computations. These improvements in performance promise to make the machine learning community take notice.

Gamalon is now announcing the first two commercial applications created using their BPS technology, Gamalon Structure and Gamalon Match. “It’s machine learning as a service,” he explains. “We can host it on any of the major cloud providers.”

Companies can use the service to structure, clean, prepare, and integrate data derived from disparate sources. Vigoda says, “More than 90% of enterprise data is unusable, because it is unstructured. Companies have large collections of little blobs of free-form text such as product descriptions, customer names and addresses, spoken queries that have been converted to text, insurance claims notes, doctor’s notes, etc. They need to convert each blob of text into a database row with the right columns. There is not really a good way to do this right now, so companies outsource to mechanical turk or professional services firms, and they get a lot of errors. Our new product, Gamalon Structure, solves this data structuring problem.”

Furthermore, if you want to link and integrate multiple data sources together, you need to use data integration or data prep software,” explains Vigoda. “You then pay ten times as much as you paid for the software to review the results and eliminate errors and redundancies. The integration takes months, and then you can have dozens of people reviewing the results.” Gamalon Match solves this data integration problem.

One Gamalon customer has hundreds of brick and mortar stores across the U.S. and wants to use Gamalon to set up an inventory system for a home delivery service. “They need to link to their stores and figure out what products are on the shelves,” says Vigoda. That may sound straightforward until you consider “how many different ways there are to describe a case of diet coke,” adds Vigoda. “Our systems goes into all the databases, reads them all, and figures out what’s available in each store, so when the driver gets to a store they know the product will be waiting.”

A manufacturing and wholesaler customer, meanwhile, “wants to know what’s going on each of its hundreds of resellers and who they’re selling to,” says Vigoda. “So we go in and connect all those databases, line them up with their contracts list, and get an incredible view of how products are moving through their distribution channel.”

Gamalon’s eventual goal is to provide “the ubiquitous middleware layer for all SaaS software,” says Vigoda. “Today, there are hundreds of different enterprise SaaS apps available, each of which stores data differently, and every company buys a different mix. We can provide a single global view of what your company is selling and who you’re selling to, where the inventory is coming from and how much you’re paying for it, all without needing to migrate to a single centralized database system. We could replace database systems of record with machine intelligence that indexes your enterprise information. We are excited to see where these first products takes us!”

About STEX25 and MIT’s Industrial Liaison Program (ILP)
STEX25 is a startup accelerator focused on fostering collaboration between MIT-connected startups and member companies of MIT’s Industrial Liaison Program (ILP). STEX25 is managed by MIT Startup Exchange, and its parent, the ILP. The ILP is a key player in making industrial connections for MIT, with over 220 of the world’s leading companies using their ILP memberships to advance research agendas at MIT.

December 5, 2016

Apps for Operators on the Factory Floor

Tulip delivers cloud services to manufacturing’s most critical assets.
Walk onto any manufacturing floor, and you’ll probably see a dramatic range in computer technologies, ranging from the latest highly automated machines to decades-old equipment with primitive data systems. Oh, and don’t forget the human operators working with pads of paper.

Each day when millions of manufacturing employees across the United States go to work on production lines, “they basically are told, put your smartphones and tablets away, forget they exist,” says Natan Linder, co-founder of Tulip. “That’s pretty crazy, and it’s something we need to fix.” Natan Linder
“In this era, we have to build apps for those people in the environments where they work,” Linder says. “Behind every product there are people, and we don't see that changing. It’s people who have the spark to figure out how to manufacture something two times cheaper, and it's people who train the next batch of employees, and it's people who have a bad day and make the terrible mistake that reduces manufacturing quality.

Tulip, an MIT Startup Exchange 25 (STEX25) firm in Somerville, MA, focuses on bringing these employees on shop floors fully into the Internet era.

The company’s manufacturing operating system allows process engineers to create shop floor apps, with interactive step-by-step work instructions that are enabled with sensing and data collection using Internet of Things (IoT) gateways. “The apps give you access through our cloud to an abundance of information and real-time analytics that can help you measure and fine-tune your manufacturing operations,” Linder says.

In a pilot project with one Industrial Liaison Program partner in biopharmaceuticals, for example, the Tulip took on the challenge of training new operators on a highly complicated, customized and regulated manufacturing process. Previously, the only way to train new operators was to walk them repeatedly through all the steps with an experienced operator and a process engineer. Tulip quickly deployed its software along with IoT gateways for the machines and devices on the process, and managed to cut training time almost by half.

App comparisons
“We enable manufacturers to create shop floor apps, much like the apps on your phone, that allow you to access data and make decisions,” Linder says. “The apps help to guide people through the manufacturing process, and to provide a host of information and analytics.”

Tulip’s operating system has three components: a layer to connect tablets or smartphones or other interfaces, an IoT layer to connect manufacturing systems and devices, and a cloud environment for creating apps and analyzing their data in real-time.

“Our cloud authoring environment basically allows you to just drag and drop and connect all the different faucets and links to create a sophisticated app in minutes, and deploy it to the floor, without writing code,” he says.

Different types of manufacturing engineers will develop different types of apps, Linder notes.

Quality engineers, for instance, typically gather data for analysis, with automated collection devices using various sensors such as thermal cameras, machine vision systems or digital calipers.

Process engineers primarily focus instead on designing work instructions. “These instructions are the key documents the industry uses to create step-by-step processes, but they usually end up as wallpaper, because humans learn and remember and just avoid using them,” Linder comments. “The problem is that this lowers the chance of changing a process, because it is tough to change a static document. It's easier to change a Tulip interactive document or app that has a digital thread and has an easy way to collect comments on the process and to share them around.”

“Once those apps are live and active, you can just hop on the backend and start querying our analytic system,” he says. “This provides tremendous insight into what you told people to do and what they actually did, with the tools and sensors and machines that they used. You have access to real-time information about what's happening on the floor, which is otherwise very hard to get, especially because you don't normally put sensors on people.”

In one pilot project for a discrete manufacturing operation, for instance, an app can show a process engineer what each individual operator carried out at a certain point in time. The engineer might see that the cycle time for one step is set to 10 seconds but nobody on the production line can hit that mark, while some operators have found a way to perform another step in the process more quickly than expected.

“That level of granularity is important when you're trying to get a manufacturing process tuned to be the best that it can be,” Linder says. “And labor is such a big component of the cost that this really comes down to the bottom line.”

Growing Tulip
A software engineer with particular interests in embedded engineering and manufacturing, Linder worked at several electronics firms before earning a PhD at the MIT Media Lab. (While at the Lab, he also cofounded Formlabs, which builds high-resolution 3D desktop printers.)

As he worked on his doctorate, Linder and Tulip co-founder Rony Kubat studied human/machine interface problems in manufacturing in partnership with Media Lab sponsors. Spending a lot of time in factories, the two graduate students spotted a much bigger problem on the shop floors.

“Operators working with very sophisticated automation were using pads of paper, which is mind-boggling,” he says. “They don't communicate or access information in the way we take for granted.”

That challenge called for a new type of information platform, and Linder and Kubat founded Tulip in 2014 to build it.

“Our ideal partners are manufacturing organizations who want to leverage their manufacturing work force,” Linder says. “They want to give tools of this century to the people on the lines, enable them to make quick decisions on the fly, and rely on them to take their production processes a step further.”

“Manufacturing often remains mass production, but products also will be built and customized to order, which means our factories need to become much, much smarter,” he adds. “They need data-driven flexible production lines, and people will be the key factor in gaining access to these new smart factories.”
About STEX25 and MIT’s Industrial Liaison Program (ILP)
STEX25 is a startup accelerator focused on fostering collaboration between MIT-connected startups and member companies of MIT’s Industrial Liaison Program (ILP). STEX25 is managed by MIT Startup Exchange, and its parent, the ILP. The ILP is a key player in making industrial connections for MIT, with over 220 of the world’s leading companies using their ILP memberships to advance research agendas at MIT.
November 30, 2016

Last Mile Delivery and Logistics Made Smarter Through Wise Systems

Tracks and optimizes delivery fleets through real time and historical data
In this postmodern world we as consumers want things faster. In fact, we want things, everything, right now. It follows that logistics and delivery systems must adapt to suit these demands. One man, Chazz Sims, CEO and co-founder of Wise Systems, is working to make that happen. According to Sims, in a culture of immediacy that includes when and how companies distribute and deliver their goods, flexible systems that respond efficiently are a necessity. At present, that is where Wise Systems is putting much of their focus: “How do we make sure our system is flexible and fast enough to do the computations necessary to make it easy for the driver to get on board with schedule changes, in order to make sure people can get what they want when they want it.” Chazz Sims
Wise Systems
Sims received his B.S. in computer science at MIT, before completing his Masters of Engineering at the MIT Media Lab. It was here, at the Media Lab, in a course called Development Ventures, that the founding members of Wise Systems first met. Sims and his partners were challenged to think about problems that affect people not just in the United States, but in a global context. This inspired Sims to better understand cities and logistics, and how both will evolve over time. He recognized that “at the heart of every city is last-mile delivery and logistics.”

The startup ecosystem at MIT functioned not only to catalyze his thoughts around last-mile delivery systems but also to provide the necessary resources and infrastructure for a fledgling team intent on designing a scalable platform that would be “elegant in its execution and flexible in its application.” In fact, it was during his participation with MIT’s highly-lauded GSFA accelerator program that he was able to interact with companies in this space, and began to practically apply his passion for utilizing data to make cities smarter. Enter Wise Systems.

Wise Systems provides enterprise software to help make businesses more efficient and deliveries more predictable. They do this by utilizing multiple data sources to track the movement of a company’s delivery fleet and analyzing real-time and historical data to optimize transportation operations. This includes data concerning the movement of drivers throughout our cities, how long it takes to complete deliveries once on-site, and variables such as customers not being present for said delivery, heavy foot traffic (think, a busy 7-11 where the intended customer can’t immediately accept a Slurpee delivery), or roadblocks related to construction.

“All of these different things seem random and unpredictable within the space,” says Sims. “So we’ve built a system that uses machine learning in order to take all of that data and build better plans for the future. Because last-mile is about planning and execution. And we believe it is about learning from those things as well.”

This forward-thinking model is what, at least in part, separates Wise Systems from the competition. Whereas other companies are spending the majority of their focus and time on the static planning of delivery logistics, under the stewardship of Sims and his cohorts, Wise's software approaches delivery logistics with a three-pronged plan of attack. First, Wise Systems helps companies automatically plan and schedule their daily deliveries under a variety of constrains. Next, they help companies execute that plan by monitoring and adjusting operations in real-time. And finally, Sims says, “We collect the data to improve the process over time…learning from what you [the client] did. Learning from historical data separates us from other companies and helps us build better technology.”

With Sims as CEO, Wise Systems has participated in a variety of startup accelerator and development programs, including Lamp Post Group’s inaugural Dynamo logistics accelerator and Techstars. They are also members of MIT’s E14 Fund, which was developed to support exceptional startups springing from the Media Lab. Sims points out the extraordinary value of participating in the MIT startup ecosystem, growing Wise System's relationships, and acquiring the necessary mentorship to help guide Wise as they connect with companies in the industry.

They’ve already worked with companies in a range of industries, including multiple global 500 companies. So far, customer response has been overwhelmingly positive, from the higher-ups, down to truck drivers on the ground. In fact, Sims and his co-founders have made a point of collecting and implementing feedback from drivers in an effort to create technology that works for everyone involved in the process—even going so far as to ride around in trucks on deliveries to get first-hand experience of what it means to use Wise Systems applications and get buy-in from the drivers. Overall, Wise makes lives easier for everyone involved in the delivery logistics process by optimizing and making the necessary adjustments.

And now, Wise Systems are newly-minted members of STEX25, which focuses on fostering collaboration between MIT-connected startups and member companies of MIT’s Industrial Liaison Program (ILP). “We’re excited to be part of the program,” Sims says. He recognizes that exposure to “…companies willing to take risks on innovation and having a network of companies excited to work with MIT…is incredibly valuable for us.”

For Chazz Sims, the future of logistics and delivery looks like a much better place, a place where goods arrive seamlessly and only when they are needed. He talks about the simple act of sitting down to eat at a restaurant, and imagines that in the not-so-distant future it will be his intelligent technology that simplifies a complicated process and plays an integral role in making sure the components of the sandwich on his (or our) plate arrived there on time and hassle free; the sandwich along with the plates, the napkins, and even the chairs. There are far reaching implications for the developing world, of course. At present, upwards of 80% of the world currently lacks access to reliable traffic data. However, with Sims as CEO, Wise looks to the future and works to make sure their platform will scale across different countries and regions. In other words, restaurant sandwiches are just a small aspect of a much bigger picture for Chazz Sims and Wise Systems.
About STEX25 and MIT’s Industrial Liaison Program (ILP)
STEX25 is a startup accelerator focused on fostering collaboration between MIT-connected startups and member companies of MIT’s Industrial Liaison Program (ILP). STEX25 is managed by MIT Startup Exchange, and its parent, the ILP. The ILP is a key player in making industrial connections for MIT, with over 220 of the world’s leading companies using their ILP memberships to advance research agendas at MIT.

November 30, 2016

Gradiant Replicates Rain Cycle for Fresh and Recylable Water

Water Treatment Startup Generates Water Quality to Spec
Anurag Bajpayee
Founder and CEO
“We like to call ourselves solving the world’s greatest water treatment challenges, which either cannot be solved with current technologies or are prohibitively expensive,” says Anurag Bajpayee, co-founder and CEO of Gradiant, a technology-driven water services company focused on treating the most contaminated industrial waste water. While Gradiant’s technologies are applicable to a wide variety of industry, its first market entry was in U.S. oil and gas. In 2012, when the shale boom was going up and water was the centerpiece discussion, both economically and environmentally, Bajpayee and his colleagues at MIT approached industry regarding the economic, logistical, and regulatory constraints in water treatment. By the end of 2013, they had developed a pilot in the field for their first technology called Carrier Gas Extraction and got the go-ahead to convert it into a commercial facility. That year, Gradiant was awarded the Global Water Intelligence “Technology Idol” award for Carrier Gas Extraction, a desalination technique for turning very contaminated water two to five times more saline than sea water into fresh water quality. Bajpayee describes it as a very simple, but creative solution. “What we were trying do is replicate the rain cycle in a confined space and in a very short period of time.”

Carrier Gas Extraction
The technique uses two unit operations: a humidifier and a bubble column dehumidifier. The water, heated by a low temperature thermal source, rises to the top of the humidifier (a tall tower filled with packing material). As the water drips down, the carrier gas (dry, ambient air) goes up, and they come in contact on the packing material, which provides a surface area for the evaporation process. The air picks up the pure water vapor and leaves behind the salts and other contaminants at the bottom as a saturated brine. By the time the air gets to the top of the humidifier, it’s pure air carrying pure vapor. “We Essentially created the cloud,” says Bajpayee. “Next, we had to create the rain.”

Perforated plates run through the inside of the tower of the multi-staged bubble column dehumidifier. Each of the stages has holes in it, and on top of each of the plates sits a shallow pool of ambient temperature or cold fresh water. As this humid air comes through the bottom of the bubble column, it passes through the holes and starts to bubble through the shallow pools of water. As that bubbling happens, a very rapid heat-transfer and mixing process occurs. The air, then, cools down and condenses the water it had picked up, because it can no longer carry the humidity. “You know how in humid air the temperature reduces and you get dew? That’s exactly what we’re getting,” says Bajpayee. “Because you’re adding more pure water to it, each of these liquid columns continues to increase in height, until the liquid column hits the overflow port and falls to the bottom where it is collected as fresh water.” By the time the air has passed the last stage of the bubble column, it’s cold and dry again, and can either be ejected or recycled in a closed loop.

Selective Chemical Extraction
At that point Gradiant’s customers said, “‘This is great, you’re actually half the cost of the existing solutions, but you’re generating pure water. We don’t always need drinking quality water. Can you give us something that’s lower cost and lower performance, as well?’ And we said, ‘Sure,’” says Bajpayee. He and his team went back to the labs and developed a second product line: Selective Chemical Extraction, a treat to speck water treatment solution, where customers tell Gradiant exactly what they need taken out. “Because if you don’t want drinking quality water, then why pay for it?” asks Bajpayee.

In both technologies one of the important things, Bajpayee noted, is the ability to handle variability. “Industrial waste water as opposed to sea water can vary quite a lot. “Most technologies are designed to work at a steady state, whereas our solutions take feed water that is changing constantly—sometimes hour to hour—and continuously optimize the system to generate product water quality that’s exactly the same every single time.”

Free Radical Disinfection
In addition to the fresh water solution and the recyclable water solution, Gradiant also has a solution that disinfects high amounts of bacteria at extremely high through put rates called Free Radical Disinfection. “This is very specific to the oil and gas industry,” he says. “As the water is going down the well for fracking or drilling purposes, it’s important to disinfect—to take the bugs out—as we say in the industry, so that it doesn’t create complications.”

With these three solutions, Gradiant now has full portfolio of water treatment and management needs of its customers. “Water is a diverse field,” says Bajpayee. “If you want to be a world-leading industrial water treatment company, you need to have a portfolio of technologies and solutions, as opposed to one silver bullet you’re trying hit everywhere.” And even though its solutions are commercial and competitive and in some cases completely revolutionary, Gradiant continues to better them with a team that is very good at taking market feedback, improving current technology, and developing new technologies and products to address other problems.

Currently commercially operating in oil and gas fields in Texas and New Mexico, Gradiant is quickly expanding into other industries including coal power plants, textile mills, leather tanneries as well as internationally. Its next generation systems of the same technology are expected to be lower cost and higher efficiency. In addition to these technologies, it is working on technology called Ion Caging, a closed loop softening system specifically focused on sulfate removals. “We are also working on directional solvent extraction, which is another desalination technology for high salinity water, focused on small scale and space-constrained applications. We are working on novel membrane systems, which promise to increase the efficiency or increase the recovery of standard sea water desalination systems,” he says.

“Technology improvement is never complete, as we know,” says Bajpayee. “At MIT we were always working on newer and better and more efficient, and that’s what we’re doing at Gradiant.” In 2014 Gradiant won the industrial water project of the year which is given to a running, working, profitable commercial project. That was the first time in the history of the organization that Gradiant went from a technology idol to an industrial project of the year globally within one year. “Successful companies and successful technologies adapt and evolve along the way. That’s where Gradiant’s strength lies,” says Bajpayee. “Our customers look at us as a good solution, but they also look at us as a long-term partner, a team that can solve any issue in water that might come their way.”
About STEX25 and MIT’s Industrial Liaison Program (ILP)
STEX25 is a startup accelerator focused on fostering collaboration between MIT-connected startups and member companies of MIT’s Industrial Liaison Program (ILP). STEX25 is managed by MIT Startup Exchange, and its parent, the ILP. The ILP is a key player in making industrial connections for MIT, with over 220 of the world’s leading companies using their ILP memberships to advance research agendas at MIT.

September 27, 2016

ILP and STEX Announce STEX25 Launch

New startup accelerator focuses on fostering startup and industry collaboration.
Karl Koster welcomes startup and industry participants at a recent STEX workshop at
MIT's Industry Meeting Center.

MIT’s Industrial Liaison Program (ILP) and its MIT Startup Exchange (STEX) are pleased to announce the launch of STEX25, a startup accelerator focused on fostering collaboration between MIT-connected startups and member companies of MIT’s Industrial Liaison Program.

STEX25 establishes an elite cohort of startup companies—25 from the 1,000+ in the STEX database—that have proved themselves with early use cases, clients, demos, and partnerships, and may be on the cusp of significant growth. These young companies are particularly well-suited for industry collaboration, and therefore will be prioritized by the ILP when advising its members on startup engagement.

The first five startups to be inducted into STEX25 are Akselos, BioBright, Lexumo Inc., Poly6, and Tagup. Each quarter six to seven additional startups will be inducted into the program until a cohort of 25 is established.

“STEX25 companies have a strong science and technology foundation in important fields such as artificial intelligence, automation, energy, healthcare, ICT, Internet of things (IoT), life science, manufacturing, materials, nanotech, sensors, and more,” said Trond Undheim who directs the Startup Exchange.

Infused with the high-caliber talent and cutting-edge technology that are the hallmarks of MIT-connected startups, these startups are poised to offer industry partners an injection of innovation and entrepreneurial spirit.

“Helping MIT-connected startups get traction with large corporate players is a crucial step in technology commercialization,” said Karl Koster, Executive Director of the ILP. “Our members are very interested in meeting with the STEX company founders, and these kinds of connections are vital to growing MIT’s innovation ecosystem.”

About STEX25 and MIT’s Industrial Liaison Program (ILP)
STEX25 is a startup accelerator focused on fostering collaboration between MIT-connected startups and member companies of MIT’s Industrial Liaison Program (ILP). STEX25 is managed by MIT Startup Exchange, and its parent, the ILP. The ILP is a key player in making industrial connections for MIT, with over 220 of the world’s leading companies using their ILP memberships to advance research agendas at MIT.
September 26, 2016

Safeguarding Open Source Software

With "Big Code" analysis, Lexumo helps Internet of Things manufacturers continuously monitor their code for security vulnerabilities.
“Open source software is eating the world,” says Brad Gaynor, co-founder and chief technology officer at Lexumo. “Products and platforms today are made up of code that wasn’t developed for a company, isn’t maintained by the company and continues to evolve independently over time. This is an entirely new paradigm in software development.”
Brad Gaynor
Open source code—created, enhanced, fixed and updated by volunteer programmers—is particularly essential in embedded devices, Gaynor points out. Smart TVs and many other everyday electronic devices, and more and more industrial control, building automation and factory automation systems, run on open source software.

Increasingly these devices also live on networks, where open source software is open game for cyberattacks. More than 8,000 new security vulnerabilities were reported publicly in open source components in 2015, according to security information provider Risk Based Security. One notable example is Heartbleed, a critical vulnerability in a commonly used communications library called OpenSSL, which was exploited by cyberattackers in the Community Health Systems breach. The attackers exploited this vulnerability in a network device and stole 4.5 million sensitive patient records.

That’s a scary situation for the makers of consumer and industrial Internet of Things products. “If their embedded open source code turns out later to be vulnerable, then brands are at risk,” Gaynor points out. “Safety is at risk. Industrial operations are at risk. Business continuity is at risk. So there’s a new requirement for these manufacturers to ensure that the devices they are putting into the world are secure.”

Lexumo, a member of the MIT Startup Exchange 25, takes on that requirement by constantly monitoring open source software for vulnerabilities and helping the companies that use open source code fix specific problems with their code. The Cambridge-based firm provides a cloud platform that systematically monitors vulnerabilities and collects best fixes for all the world’s open source software.

Open source, open risks
In today’s manufacturing firms, “programmers don’t write code from scratch anymore,” Gaynor says. “They get the pieces of open source code they need, they stitch it together, they make it their own, they build their products out of it, and they might add an algorithm as their secret sauce. They wrap their brand around the outside of the product, and they race it off to market.”“But the open source code they’ve used continues to evolve, and it may become vulnerable,” he says. “The trouble is, by the time a company has made this code their own, it’s really difficult to become aware that there’s this new problem, which comes up out of the blue, with no tie to the product lifecycle. Companies need to respond quickly to secure their products. They need to incorporate the fix and the upgrade without breaking everything they’ve done. Therein lies the challenge.”

A few years ago, while Gaynor was managing cybersecurity projects at the Draper Laboratory, he and his Lexumo co-founders came up with an idea on how to overcome the challenge. “We could take all the world’s open source software, every package, every version, and develop a semantically searchable library of all code,” he says. “We then could look across all software programs and learn something about the whole.”

“This concept is called Big Code—Big Data meets software analytics,” Gaynor remarks. “Big Code is a new domain, and Lexumo is the first commercial instance of Big Code. We have built the first platform to look at all code to analyze some code, and that allows us to do things at scale that have never been even tried before.”

He and his colleagues painstakingly developed an initial proof of concept for their approach in research carried out at Draper for the Defense Advanced Research Projects Agency (DARPA). “Open source security is a very tough challenge, and we realized that that the commercial world needed this as much as the government sector,” he says. Draper provided seed funding for the startup company, and Gaynor and co-founders Nathan Shnidman (an MIT alum) and Richard Carback III launched Lexumo in 2015.

Big Code for better code
Lexumo’s platform had to overcome three daunting technical barriers: Gathering, analyzing, and annotating open source software vulnerabilities.

“Getting all the world’s open source software was probably the easiest of the three challenges,” Gaynor remarks. Modern software repositories essentially store all of the code, in all of its versions.

But searching and analyzing all this code, which is written in many programming languages, was a much tougher problem. Lexumo has tapped two major software innovations to solve the problem.

One innovation was modern NoSQL graph databases that make all this data searchable at scale. “We’re talking about terabytes of data just to get started,” Gaynor notes.

The other key innovation was in software compilers—programs that transform code from one language to another and are typically used to create executable programs. Lexumo tapped into the LLVM compiler project, which offers an infrastructure that can take almost any software language into a common intermediate format. “By bringing any language into a common intermediate format, for the first time we can look at all code in a common representation,” Gaynor explains. “So we can use a set of analytics that can look at all open source software as if it has a common language.”

The third major challenge for the Lexumo platform was annotating the code with the collective human knowledge about vulnerabilities affecting open source. “This turns out to be a new trick, because vulnerabilities are disclosed by people, and this information that people contribute lives in many different places and many different forms,” he says. “But when it’s put into the vulnerability databases, it’s often wrong. The records tend to be incomplete and self-inconsistent.”

However, with Lexumo’s ability to search all of the world’s open source software, “we can look at each one of those security disclosures as a tip,” he says. “We’ve developed machine learning algorithms that allow us to very quickly, with minimal human intervention, identify exactly what the flaw is that the disclosure reflects and exactly which versions of the code have that flaw.”

With this capability, by the time the platform alerts customers about a risky piece of code in a specific product, “we know by lookup exactly which vulnerabilities affect them, and we can determine exactly how the open source community remediated the vulnerability,” he says. “We can extract that information as a patch that we provide to our customers. They get the alert from us that this problem exists, and here’s how to fix it.”

As a cloud-based platform, Lexumo can take a light touch in aiding software developers, working smoothly with their existing tools. The best way to communicate the platform’s value to potential clients is simply to demonstrate the software on one of their projects, Gaynor says. “In many cases with large companies, we can run a product through the platform and demonstrate exactly the results they can expect,” he says. “We generally have new customers up and running within half an hour.”

Demand for such security services can only climb, Gaynor says. For instance, imagine the risks that your latest big screen TV might bring. “The TV is a network device and it probably has a camera,” he points out. “If there is a software vulnerability in that networked camera, an attacker might come into the camera, turn it on without even turning on the TV, and start to pipe out video from your living room over the Internet. Similarly, cyberattackers might exploit software vulnerabilities to tamper with industrial IoT systems such as smart buildings, factory automation or power generation.”

“This is why commercial vendors of the Internet of Things are worried about their brand and why they invest in security,” he adds. “As the world goes open source, the way we solve security problems has to change, and that’s what Lexumo does.”

About STEX25 and MIT’s Industrial Liaison Program (ILP)
STEX25 is a startup accelerator focused on fostering collaboration between MIT-connected startups and member companies of MIT’s Industrial Liaison Program (ILP). STEX25 is managed by MIT Startup Exchange, and its parent, the ILP. The ILP is a key player in making industrial connections for MIT, with over 220 of the world’s leading companies using their ILP memberships to advance research agendas at MIT.
September 26, 2016

Industrial Equipment Meets the Cloud

Tagup’s platform tracks equipment operation and avoids unplanned downtime.
On Jon Garrity’s iPhone, Tagup is displaying a power transformer at a nearby power plant: not just photos of the transformer, but real-time data on the equipment’s operations such as the temperature and dissolved gas level of its oil. If there were any clues here of something just starting to go wrong, an expert might spot them immediately.

Jon Garrity
That’s the core point of the Tagup platform, which connects industrial equipment to cloud-based services. Tagup provides remote visibility of the equipment, along with better understanding and predictions of its performance, to operators, third-party experts and manufacturers.

“We help reduce unplanned downtime, which has huge costs in lost output and productivity,” says Garrity, co-founder and CEO of Tagup, a member of the MIT Startup Exchange 25 based in Somerville, Massachusetts. “Monitoring data continuously coming off this equipment, an operator or third-party expert can identify an issue before it occurs.”

These benefits are compounded by the ability to generate better operational models by analyzing industrial equipment at scale. “We’re collecting a much higher-resolution set of data, and we can collect it across a much larger sample size,” says Garrity. His co-founder (and fellow MIT alum) Will Vega-Brown, an expert in machine language and software, is leading the company’s efforts to exploit these analytic opportunities.Watching from the cloud
Current monitoring systems for high-cost machinery such as gas turbines offer a glimpse of what Tagup delivers more broadly. In one example, General Electric sells a gas turbine with an accompanying remote monitoring solution. A GE network operating center can watch the performance of more than 1,700 gas turbines around the world, with a team that identifies any problems starting to crop up and then helps operators deal with the problem proactively.

“Our platform enables that type of service for any equipment type,” Garrity says. “If you’re operating, say, a power transformer, instead of relying on once- or twice-a-year oil sampling, you can have third-party experts monitoring the data continuously, with computer analytics, to help you cut downtime.”

As industrial machines join the Internet of Things, Tagup aims to fill a huge and growing gap. “Currently there is no way to simply remotely monitor a diverse number of equipment types and share that data with third parties,” Garrity says.

One challenge is that different pieces of equipment work on different data and communication protocols, some of which have been around since the 1960s, he explains. Tagup translates the data from these older protocols into a modern web protocol, so that developers can build the tools and the analytics required to enhance, modernize and optimize equipment operation in the field.

Sometimes this installed equipment can make its operating data accessible remotely, sometimes not. “If not, we connect the equipment directly with an off-the-shelf industrial cellular gateway—basically a rugged cell phone,” he says.

Making the connection is typically straightforward—getting information about the equipment and the data it provides to provide a “digital twin” and then going onsite to hook up the gateway, which can take as little as 15 minutes. “The rest we do remotely,” Garrity says. “Data coming off that equipment is sent securely to our cloud, where we parse, store and analyze that data, and then make it accessible to users through our application. Our customers are looking for turnkey value, and that’s what we provide.”

Proactive help for steady production
Among early Tagup installations are reverse osmosis systems that deliver purified water. These systems use membranes, whose performance degrades over time. “With Tagup, you can remotely monitor how efficient your membranes are, and you can see when they need to be replaced,” Garrity says.

One customer, a value-added reseller for reverse osmosis systems, can see in real-time the health of all equipment that they have installed, and prioritize its maintenance operations. Instead of visiting each plant every three months, the reseller can focus on locations where maintenance issues are coming up. And it can do so across various types of reverse-osmosis facilities. “They’re using one software interface to check on the health of all the equipment they’ve installed and serviced to date,” Garrity says.

Additionally, the Tagup platform can offer major payoffs for equipment manufacturers. “They can know exactly who the end customer is, exactly where the product is and exactly how well that equipment is performing,” he says. “There’s also a new sales and service opportunity presented to them, moving from a pure equipment sale to a higher-margin service business.”

Moreover, the value of equipment monitoring increases with scale, which allows the generation of better models for predicting equipment health and performance. “The results from our machine language analytics are not yet established, but we’re optimistic about their capabilities,” Garrity says.

For example, in a power transformer, how does the oil’s temperature correlate with a given maintenance concern? “We can build models that take into account, say, 30 of these variables, looking across thousands of transformers, to help identify problems that wouldn’t be picked up by existing models,” he says. “The more data we integrate, the better analytical models we can make, and we can just push these models to the cloud.”

“Given the size of the industrial Internet of Things, the market for these applications is enormous,” Garrity points out. “For instance, sales of new power transformers make up about a $6.5 billion annual market, on top of a large installed base. Analyzing the operational data via our platform and helping customers cut their unplanned downtime creates real value.”
About STEX25 and MIT’s Industrial Liaison Program (ILP)
STEX25 is a startup accelerator focused on fostering collaboration between MIT-connected startups and member companies of MIT’s Industrial Liaison Program (ILP). STEX25 is managed by MIT Startup Exchange, and its parent, the ILP. The ILP is a key player in making industrial connections for MIT, with over 220 of the world’s leading companies using their ILP memberships to advance research agendas at MIT.
September 26, 2016

Fixing the Lab Reproducibility Crisis with Augmentation, not Automation

BioBright strives to augment human ability to do science in the lab with voice assistance that recognizes biomedical research terms and the use of augmented reality tools.
A recent Nature survey of 1,576 researchers found that more than 70 percent of respondents have tried and failed to reproduce another scientist's experiments. More than half have failed to reproduce their own. This “reproducibility crisis” was revealed to be especially acute in biological and medical research.
Charles Fracchia
“Lack of reproducibility is a big issue right now in biomedical research,” says Charles Fracchia, CEO and founder of MIT-based startup called BioBright, which is aiming to improve laboratory documentation technology. “More often than not, it’s very difficult to determine the root cause of an experiment’s outcome. It’s costing us between $10 and $50 billion dollars a year in the U.S. alone.”

Most of the Nature survey respondents agreed that the solution to the crisis is improved documentation and standardization of protocols. Yet given time pressures, few were willing to put in an estimated 30 percent more time required for comprehensive documentation.

“Current laboratory tools are not conducive to reproducing experiments,” says Fracchia, a bioelectrical engineer who is currently on leave from the MIT Media Lab. “Most tools don’t connect to networks or have outdated connection mechanisms and proprietary formats. All that stands in the way of doing longitudinal, data driven discovery.”
To ease the documentation process, Cambridge, Mass. based BioBright offers a suite of “smart lab” software, hardware, and services. The company built a voice assistant called Darwin tailored to recognize biomedical research terms, and software that automatically collects data from laboratory equipment. Information is recorded, aggregated, and analyzed in the cloud where other scientists can access this integrated record in order to duplicate the experiment.

“Our goal is to augment the human ability to do science in the lab with voice assistance and augmented reality tools,” says Fracchia. “With BioBright, scientists can spend their time uncovering the root causes of success or failure while staying in the experiment. They don’t have to stop every 15 minutes to record information. Our system allows the lab notebook to write itself so scientists can do what they’re supposed to do -- analyze the data.”

BioBright is targeting biomedical researchers in academia, biotech, pharma, and even healthcare, and the tools could eventually expand to other industries such as food research. The company was recently chosen to be among the first six companies in the new MIT Startup Exchange 25 (STEX25) program. This more focused version of the successful MIT STEX program is designed to facilitate industry interaction with the 25 most promising MIT-based startups.

Augmentation over Automation
Some have argued that the reproducibility crisis can be solved with laboratory automation -- replacing human-driven lab processes with robots and other devices. Fracchia himself helped launch an earlier MIT-based startup named Gingko BioWorks that built the first automation tools for synthetic biology.

The experience taught him that while automation is often valuable, it does not necessarily improve reproducibility. “Automation is excellent at doing workflows and protocols 1,000 times over, but it’s very brittle, and has no ability to adapt,” says Fracchia. “Research is for the most part an exploration. You’re adapting and changing parameters on the fly. BioBright’s augmentation tools keep humans at the center of the loop, not computers.”

Fracchia goes on to note that the automation world is mostly driven by computer scientists “who don’t know the intimate problems of the lab.” BioBright inverted that. “We are biologists with intimate knowledge of biomedical workflow who learned computer science and electronic engineering.”

According to Fracchia, most reproducibility problems stem from translating workflow between researchers. “Automation won’t help with that,” he says. “BioBright’s human augmentation, however, allows the scientist to say ‘Darwin, show me the average temperature that I’ve used for the last three months, or show me how Mike did it last week.”

BioBright helps reduce the often significant time lapse between data generation and collection, says Fracchia. “Researchers’ hands are often busy and gloved, so they must put off the time when they record information, sometimes even to the end of the day,” he says. “A lot of information gets lost this way. Small deviations from the protocol are often not recorded. Maybe you will note on a Post-It that a sample was a little more viscous than usual, or if you’re really disciplined, write it in a notebook, but there it stays. It’s difficult to look at the information in context.”

BioBright helps solve a related challenge with laboratory documentation: the vast range of time scales. “In biological research you’re ranging from picoseconds to hours, days, or even months,” says Fracchia. “Humans are good at collecting information in the minute range, but have difficulty in other ranges, especially picoseconds. BioBright pervasively collects information across different time scales and centralizes it in one place.”

Inside BioBright – from Custom Sensors to Voice Control
The sensors and cameras available with BioBright vary depending on the specific environment, but the underlying architecture remains the same. BioBright is a cloud-based platform that uses a modest onsite computer – currently a Raspberry Pi board – to act as an Internet of Things hub. The local device aggregates information from the lab sensor network and mediates with the cloud service using end to end encryption.

One of BioBright’s key innovations is a “hot folder” that interfaces directly with lab equipment. “We automatically grab the data as it’s generated and centralize it,” says Fracchia. “The system is built to be extensible and include new data formats based on customer’s needs, which allows us to extract metadata that is directly relevant to our customers’ workflows.

BioBright offers camera systems that record in both visible and infrared light. “You can place these cameras around your lab, or over benches or specific stations,” says Fracchia.

The company has also developed tiny sensors designed for biological research that can fit into standard sized vessels or tubes. “We developed the first temperature sensor that fits in an Eppendorf tube,” says Fracchia. “Our wireless sensor lets you easily record the temperature of your samples across an experiment.”

BioBright’s Darwin voice assistant enables researchers to issue voice notes instead of stopping to record information manually. “You add a little microphone to your lapel so you can interact with BioBright, and leave voice notes,” says Fracchia. “You can also give orders like telling the camera to record an image.”

The natural language AI system also works in reverse, letting you ask the computer for information or to correlate data. You can even set up the system to volunteer advice based on sensor input and historical data.

“BioBright is bidirectional, sending longitudinal information back to the scientist,” says Fracchia. “For example, we can warn scientists that they’re conducting a test at the wrong temperature. Even if they ignore the warning, they can analyze the difference between what actually happened and what was supposed to happen. At an early stage, we can tell you that your protocol is unlikely to succeed, which is tremendously important in pharma and biotech where it can take weeks to get results.”

BioBright is built on modular components that the company assembles for customized services sold to large industries. Some components, however, will be sold as standalone products.

Although most of BioBright’s technology is proprietary, Fracchia is a proponent of open standards and interoperability, which are often lacking in the biomedical field. “We use interoperable data formats so you’re not trapped in the ecosystem,” he says. “BioBright solves a problem that is common across a number of industries: scattered data in different formats.”

The company is now working on integrating sensor networks, wearable sensors, and augmented reality head mounts into the system. Eventually, Fracchia envisions something like an Iron Man suit for biomedical researchers. “We want BioBright to fit the workflow of the scientist like a glove.”

Impact before Income
BioBright draws extensively on MIT Media Lab’s innovative pervasive computing technology. Fracchia mentions the Media Lab’s Principal Research Scientist Shuguang Zhang as being especially helpful in launching the company. Other MIT institutions have also played a big role.

“I cannot say enough good things about the Venture Mentoring Service, which has been instrumental in getting BioBright to where it is today,” says Fracchia. “And MIT’s Technology Licensing Office helped us look at innovation in a refreshing way. They really understand MIT’s motto: Impact before income.”

That motto has steered BioBright away from venture capital for the time being. After raising a modest angel round, the company has been sustained entirely by customer contracts, which Fracchia says is quite unusual. “A four-year return rate was not the best match for us,” he says. “We don’t have to make investors happy or report to a board that is driven around valuation. Instead, we can focus on partnerships and analyzing companies’ reproducibility problems. We want to solve problems, not just sell you a product and go away.”

Fracchia says his team is “very honored and humbled” to be named to MIT’s STEX25 program. “STEX and the ILP have let us grow and better understand our customers,” he says. “The point of STEX25 is to link up transformative technologies with companies that are creating real value.”

About STEX25 and MIT’s Industrial Liaison Program (ILP)
STEX25 is a startup accelerator focused on fostering collaboration between MIT-connected startups and member companies of MIT’s Industrial Liaison Program (ILP). STEX25 is managed by MIT Startup Exchange, and its parent, the ILP. The ILP is a key player in making industrial connections for MIT, with over 220 of the world’s leading companies using their ILP memberships to advance research agendas at MIT.