Updated below with video from event.
The Summer 2016 cohort of startups at Microsoft Accelerator’s Machine Learning Demo Day in Seattle on Thursday revealed innovations that could save lives, make applying to college easier and save retailers time and money — and that was just the start.
“Startups are super important for our future,” said Tzahi (Zack) Weisfeld, the general manager of Microsoft Accelerator. “It’s all about picking the best startups and helping them be successful.”
Microsoft Accelerator has been mentoring nine Seattle startups for the past four months, aiming to help them create the next big thing in tech. Through the program, emerging entrepreneurs gain access to business mentors, tech and marketing experts as well as various resources to keep their heads, literally, in the cloud.
Some of the startup founders credit cloud computing as the key factor in realizing their visions.
“For decades, people have been talking about building an A.I. doctor. Now you have all of this medical data that is electronic through cloud computing. You really have the data to teach a computer medicine. That’s why the time is now,” said Arturo Devesa, the CEO and founder of MedWhat, an artificial intelligence virtual medical assistant.
The MedWhat medical A.I. sounds a lot like the star of Disney’s Big Hero 6, Baymax, a medical robot. It provides personal attention to medical and health needs by accessing a person’s profile, which includes health and medical records.
“When I want to find out if I have a cold, I go online and I see a bunch of links,” Devesa said. “But, when I click on these links, I feel like I’m reading a novel. None of these links is hooked up to any of my medical content. You need your medical context. And no search engine follows up. These are problems that we want to solve by using a medical chatbot.”
MedWhat’s chatbot essentially forms a relationship with users by accessing health and medical records. It uses that data to provide information based on patient-provided symptoms and history. For example, if the user says that they have a cold, the chatbot will ask for symptoms, like whether they have a fever, then give them information on whether they have a cold and the best ways to treat it. They can also use the information provided in medical records to intuit possible medical issues, like a user’s body mass index determining why they only sleep a few hours a night or that the patient has only slept two hours one night being the possible cause of a headache.
Devesa stressed that the chatbot does not diagnose nor make medical conclusions but is meant to provide users with a “learning and discovery process.”
“The product itself is really practical. You’re applying A.I. to a real practical problem that can create a real benefit to society,” he said. “We’re at the beginning of the fourth industrial revolution, which is A.I. We feel that medicine will be one of the biggest impacts for A.I. because it’s something that relates to everyone.”
Another startup that caught our attention was OneBridge, which predicts oil and gas pipeline failures in an effort to save lives and protect the environment. Brandon Taylor, the CTO of OneBridge, likewise calls the cloud a “game changer.”
“The cloud actually opens up the capability of building a better infrastructure because we can scale on a computer. Machine learning makes it so that you can anticipate any problems,” he said. “It’s really hard for a human to go down and spot patterns. We couldn’t do it without the Cloud and machine learning.”
OneBridge’s presentation started with a sobering visual of a pipeline explosion amidst a cluster of homes accompanied by the 911 call. Taylor’s reasoning for OneBridge’s necessity reads like a laundry list of environmental and economic catastrophes: “Every year over the last 20 years, on average, the (oil and gas) industry spills about 500,000 gallons, 20 people are either injured or killed annually, costs have increased to about $3.6 billion, and over half of the pipelines in the ground today are past their useful life.”
Prevention is antiquated and expensive, he believes, with engineers pouring over Excel spreadsheets that are “like finding a needle in a haystack,” according to the presentation. Inspections rely on odometer readings from devices – called “pigs” – that are stuck into pipes every five years.
“The industry spends almost $8 billion annually on corrosion detection,” Taylor said. “The problem traditionally has been that there hasn’t been the technology. OneBridge helps the pipeline operator detect and calculate corrosion growth for them, which helps reduce their risks and costs.”
OneBridge isn’t intended to replace oil and gas engineers. Instead, it would enhance their work by ingesting data with a user experience to create analytical dashboards to do diagnostics. Microsoft HoloLens devices in the field would allow for a deeper level of analysis on pipes, providing more information on pipeline corrosion.
“There’s 12.5 million U.S. households within the impact radius of a pipeline failure; that’s near a pipeline,” Taylor said. “This is also super important to our economy. Everything we touch every day, plastic and golf balls, is from this huge natural resource. (OneBridge) can make it a smarter infrastructure, so we can preserve it and make it last longer. We really view it as a smart infrastructure plan.”
After the presentations, some audience members said that they were most interested in the marketing intelligence platform Affinio. Surrounded by visuals of cheering crowds, CEO Tim Burke detailed how Affinio uses a graph based on consumers’ interests to help better understand market demands and values.
“We’re transforming the way marketers understand and relate to their customers,” he said. “Imagine a world that’s full of caring and relevance. That’s the world Affinio is creating.”
Here are the other startups from this Microsoft Accelerator cohort:
• Agolo, which scans, organizes and summarizes news, documents and enterprise data. “We believe that summarization is the future of search because everyone wants to be the smartest person in the room,” the company says.
• simMachines, which finds complex patterns in large data sets, and uses those patterns to make recommendations and predictions.
• DefinedCrowd, which uses crowdsourcing and machine learning to process and create structured data for scientists.
• Knomos, an enhanced knowledge management platform for corporate legal departments, leveraging data visualization and machine learning technology.
Plexuss helps uses student profiles to help them find and communicate with prospective colleges. “Plexuss is radically changing the way that students and colleges find and interact with each other,” the company says.