Algorithms are all around us, with both seen and unseen effects. Researchers are using automated intelligence to power predictive technology in everything from self-driving vehicles to climate models. But while much has been made of AI’s broader effect on society, it is already affecting the ways we live and work in very tangible ways — from helping allocate a marketer’s budget by day to determining his ideal sleeping temperature by night.
AI isn’t the only driver of automation. “Always on” technology — not quite AI, but automation tech nonetheless — helps drive investment portfolios for hundreds of thousands of people all day long.
Here, we take a look at how automation is touching various facets of daily life, from delivering a user’s sleep metrics in the morning to helping her with procurement services in her corporate legal department job. Her grandfather’s cancer is being diagnosed and treated by doctors using AI, while her therapist’s work is informed by AI-analyzed transcripts of conversations. At day’s end, digital tech is distributing her savings among a socially responsible investment portfolio and algorithms are serving up content in anticipation of her browsing her favorite e-commerce site. It all adds up to a more streamlined and optimized experience of daily life, with more space and time to devote toward the kind of work, leisure and creativity humans are really good at.
Much of this very futuristic world is being built right here in New York City — here are six companies doing just that.
A Better Night’s Sleep
Eight Sleep leverages machine learning to optimize the 26 years (on average) that we spend asleep, with a smart mattress that tracks users’ tosses and turns, body temperature and total sleep time. The technology controls the mattress’ temperature by flowing cool or warm water through a thin layer within the mattress, which is also filled with IoT sensors like piezo films, ballistocardiographs and temperature sensors. These can detect a user’s heartbeat and breathing patterns.
Between its product launch in December 2016 and June 2018, the company recorded a cumulative total of 2.5 million nights of sleep, all of which train algorithms to determine an ideal mattress temperature for each individual. As Head of AI and Research Manish Chabliani told Built In NYC in February, “Our models analyze sleep biometrics and data to predict presence in bed, detect sleep onset and to classify sleep stages.”
Our AI intelligence can tell you things like ‘Your percentage of REM sleep was 15 percent last night.’”
While regulating temperature at the individual level, Eight Sleep also delivers metrics to users on how their sleep compares to others. A user’s REM sleep, movements and delay between going to bed and falling asleep can all be tracked, and used to identify potential periods of stress. Those insights come in the form of an app-based “sleep coach.”
“Our AI intelligence can tell you things like ‘Your percentage of REM sleep was 15 percent last night. This is lower than the healthy range of 20-25 percent. Try sleeping 30 minutes more tonight,’” co-founder and CEO Matteo Franceschetti wrote in a blog post. Another example: “‘Great job! You got 7 hours and 43 minutes of sleep last night. That is 30 minutes more than your average this month.’”
Opening the Legal Services Market
Bodhala co-founders Raj Goyle and Ketan Jhaveri saw a “broken market” in legal services: rising service rates that had continuously outpaced rent, college tuitions, health care premiums, inflation and other costs that might contribute to escalating legal costs. The bottom line: law firms were acting according to the rules of a non-competitive market, and simply inflating prices because they could.
The company’s VP of Strategy, Chris Bennett, laid out the state of the industry in a recent blog post, and explained how it inspired Bodhala’s creation. The idea behind the company is to introduce competition to the industry by circumventing the “enormous pricing power of law firms” and create a “source of truth — a comprehensive dataset that gives the demand side an understanding of how law firms are price-takers in the market.”
This trove of information makes a general counsel’s team much more effective at legal spend management.”
This “source of truth” is known as Hercules, Bodhala’s proprietary data science platform that automates the legal procurement process, analyzing rates, practice areas, client types and other data points for any given law firm. Hercules then compares those services to the wider legal market. Bodhala says its platform puts more power in the hands of corporate legal departments, and makes for a more transparent budgeting and procurement process all around.
“Our proprietary benchmarking metrics and rate review algorithms generate detailed insights into every aspect of legal spend,” Bennett wrote. “This trove of information makes a general counsel’s team much more effective at legal spend management and getting the best outcomes from that spend.”
Give My Money To Your Robots
Few industries better demonstrate the complementary potential of combining automation with human expertise than wealth management. In a nutshell, robots manage the day-to-day transactions that, taken together, comprise an investment strategy — monitoring price changes, spreading capital across a range of vehicles to minimize exposure to any single financial instrument. Meanwhile, human experts oversee the strategy behind those trades, building portfolios and the technology to enact them.
Fintech company Betterment is one of several companies that has built a business from this arrangement, with a staff of traders, tax experts, quantitative researchers and behavioral scientists, among others. In a recent blog post, Quantitative Investing Associate Sebastian Rollén described technology as his team’s “force multiplier.”
Machines are generally better than people at rule-based decisions, calculations at scale and data aggregation.”
“For example, while our tax-loss harvesting algorithm was developed by the experts who work on our investing and trading teams, it would be impossible for us to manually harvest losses for our customers,” he wrote. “We’d have to monitor the daily balances of hundreds of thousands of accounts simultaneously.”
Despite the emergence of automation in investing, Rollén said humans will retain a critical role in the long term.
“Machines are generally better than people at rule-based decisions, calculations at scale and data aggregation,” he wrote. “People are usually better at complex decisions, abstract thoughts and flexibility in logic and inputs.”
Automated Cancer Diagnosis
As medical information moves onto digital servers, artificial intelligence algorithms are able to comprehend the data at scale and learn things that humans cannot. One example comes from Paige, which has built machine learning technology that learns from visual and clinical data. The idea is to help pathologists reach their diagnosis faster. Through a partnership with Memorial Sloan Kettering Cancer Center, the company has access to some 25 million pathology slides — one of the largest such archives in the world — to train its algorithms. Paige says it has enabled pathologists to find new connections between pathology, treatment response, genomics and a patient’s overall outcome.
“This area of medicine is still very microscope based, it’s very analog,” CEO Leo Grady told Built In NYC when the company announced another $20 million in funding in July. “Now, with this new technology — both for digitizing the slides to scale and pattern matching to scale — it allows us to really start providing new information to pathologists up and down the cancer spectrum, to help them do their job more effectively, and improve the outcomes for patients.”
I think we can substantially and strongly impact the current practice of pathology and, consequently, cancer care.”
Grady believes Paige has a large role to play in the evolution of pathology and cancer care as hospitals move their everyday workflows onto digital platforms. Even drug development teams can use Paige to help expedite their go-to-market processes.
“I think we can substantially and strongly impact the current practice of pathology and, consequently, cancer care,” Grady said. “We’re at the cusp of a lot of different technologies, whether it’s cloud, cheap storage or digital slide scanning, or AI. All of these things are coming together in a way that is making this sort of technology possible for the first time.”
Amid the current wave of technology allowing us to work, learn, consult doctors and catch up with friends from home, therapy is undergoing its own digital revolution, too. For example, New York-based Talkspace connects customers with therapists via chatroom, text message, audio messages and video.
The company is also bringing an element of automation to what is traditionally a highly personalized field. Talkspace is applying AI algorithms to anonymized therapy session transcripts to identify trends in user behavior, speech patterns that might signal avenues for treatment and possible early warnings of larger issues.
It’s pretty surprising that we use language in a very similar manner when we deteriorate.”
“It’s pretty surprising that we use language in a very similar manner when we deteriorate,” co-founder and CEO Oren Frank told Bloomberg in May.
The company’s technology accounts for the stage of therapy a person is at, and their relative level of engagement with a counselor to help identify a potential crisis. Meanwhile, data scientists can use insights they find to assist the company’s staff of therapists in their interactions with patients.
Anticipate Customer Intent
For e-commerce brands that live or die on the margins between sales revenue and cost-per-acquisition metrics, the ability to predict customer intent is key. Leveraging data to predict whether a visitor to an e-commerce platform will convert — and for how much — helps businesses assign marketing budgets in split-second ad auctions. Technology developed by New York-based <intent> plugs into e-commerce platforms to make real-time predictions about the likely next steps of individual users, leveraging AI on huge behavioral datasets.
“We build machine-learning models that predict the value of every individual user in real time, using only a retailer’s first-party data,” Senior Product Manager Sinan Zhang told Built In NYC in February. That data informs user acquisition metrics, onsite personalization engines and overall business forecasts.
Rather than fighting only to win conversions, [successful brands] see fragmented shopping as an opportunity.”
In a blog post focusing on the travel industry, a big customer segment for <intent>, VP of Global Marketing Susan Billingsley highlighted how a singular focus on customer conversions can be counterproductive. With users shopping across multiple platforms and abandoning digital carts filled with goods before purchasing, she encourages brands to double down on the media side of their business, too.
“Rather than fighting only to win conversions, [successful brands] see fragmented shopping as an opportunity,” she wrote. “They capitalize on user behavior and delight customers wherever they are in their journey. Sometimes that means guiding a user that’s ready to buy a conversion. Other times, it means providing effective media to a customer still shopping around.”
The effect is to create value from non-converting customers — a business model made possible by technology that can automatically predict why a user is on a platform in the first place.