machine learning and the IoT

How 3 Companies Are Using IoT and Machine Learning to Change an Industry

Welcome, IoT Day readers!

If you like this post, you may also like…

• Value of Data: The Dawn of the Data Marketplace

• The IoT Evolution – Top 9 IoT Use Cases

• Untapped Data – Getting More from IoT Data at the Edge and in the Cloud

• Are IoT-Enabled Locks the Future of Remote Security?

For years, machine learning and the Internet of Things have promised to maximize data value across industries.

But despite hypothetical use cases like sensor-laden smart roads1, successful implementations of these technologies often remain elusive.

Push past the headlines, however, and you’ll find instances where machine learning and IoT have found a place in production in a wide range of industries. And some of them are being used in unexpected and highly profitable ways, delivering on the oft-touted promise that these technologies will reshape the way businesses operate in the future.

machine learning reinventing auto insurance

Reinventing Auto Insurance

For insurance providers, questions about driver coverage are long-standing and numerous: Should a good driver have to pay as much as a questionable one? If a car is involved in an accident, since both sides are likely to point the finger, how do you know who was really at fault?

The founders of Rome-based Octo Telematics ,2 looked to onboard sensors to answer these questions. And with the rise of IoT and machine learning technology in recent years, those answers are finally becoming a reality.

“Octo was founded to create a world-class telematics solution for the insurance industry,” says Paolo Cellini, Octo’s VP of Business Innovation and Tech Scouting. “Our goal is to reduce the cost of insurance by reducing the cost of claims. The insurer saves money, and they pass the savings on to the customer.”

Octo partners with insurers as well as companies with vehicle fleets (including major rental companies and car sharing businesses) to outfit cars with a box of Octo-designed sensors that attaches to a car’s OBD (on board diagnostics) port, or near the battery in cars that lack OBD.

“The sensors can detect acceleration, location, and—most importantly—whether there’s been an accident,” says Cellini. “Every relevant type of data” is pulled into Octo’s machine learning models, which forecasts driving habits and improves insurers’ response after a collision.

When armed with this information, it’s much easier to assign fault in a collision. Sensor firmware can be updated over the air, and drivers who opt-in can access a dashboard that lets them know how well they’re driving—and potentially receive discounts on their insurance based on Octo’s ML-driven models of how risky their driving behavior is. Moreover, research suggests that by 2030, 2.5 million accidents will be prevented through machine learning and other safe-driving technology like edge computing, amounting in $22 billion in avoided insurance claims for car repairs and medical bills.

Saving the Bees

If you’ve read science headlines lately, you’ve probably learned about the collapse of bee colonies over the last few years — and the effects a lack of bees can have on agriculture (among other things). As a result, beekeepers are more concerned about the health of their hives than ever.

machine learning saving the bees

Here to provide the monitoring solution bee farmers need? ApisProtect ,3 an Irish startup founded by Dr. Fiona Edwards Murphy, an electronic and electrical engineer with an intense interest in bees. The goal of the company is simple:

“We use the internet of things to help beekeepers reduce losses and improve the productivity of their hives,” she says.

The ApisProtect hardware is a box of five sensors that resides within each hive. The sensors measure environmental conditions like temperature, humidity, and sound, relaying this data wirelessly to the company. ApisProtect then applies machine learning techniques to the data to feed a report back to the beekeeper, converting raw numbers into actionable information. For example, the ApisProtect model looks for statistical patterns in the environmental data that might indicate that the queen bee has died or that the colony is preparing to swarm.

The report may suggest the keeper heat or cool a hive, or it may suggest a certain disease treatment be applied. While still in development and on track for wide release later this year, the technology is already installed in 200 hives across 20 locations, collectively monitoring 9 million honeybees.

Streamlining the Supply Chain

Major retailers, who rely on the increasingly global world of shipping and logistics, can have billions of dollars of inventory tied up in overseas manufacturing facilities, on container ships or lingering in customs at any given time. For example, for one global fashion retailer, logistical problems contributed to an inventory glut amounting to $4.3 billion in mid-2018.

machine learning streamlining the supply chain

Understanding where products are in the shipping pipeline is a key solution to this, says Adam Compain, CEO of ClearMetal ,5 which is leveraging sensors and machine learning to give importers a better handle on the whereabouts of their goods.

“Data in the transportation world is fundamentally flawed,” says Compain. “Various systems will say a product left China at three different times. Another system will say a product is at the factory and out for delivery at the same time.”

To deal with this challenge, retailers often buy more than they need in order to avoid stockouts – and they end up often having to immediately mark down the excess inventory when it arrives. ClearMetal’s data science techniques take a variety of data sources, both public and private, including everything from ship-based sensors to weather reports, to turn all that data into something more reliable and useful, intelligently determining with much greater accuracy where a product actually is in the supply chain. ClearMetal’s machine learning platform is designed to determine which data sources are the most reliable and the most relevant for any given customer, building a location model that is structured around the very best information, downgrading the chaff.

Compain says ClearMetal significantly reduces the error rate in freight data, and it has decreased the average time a shipment spends idle after arriving in port—reducing the need for buffer stock and improving cash flow all around.

Machine Learning on the Rise

While machine learning applications still remain elusive for some, these companies are among a growing number of businesses across industries that are harnessing machine learning.

In fact, experts predict machine learning6 will be one of the fastest-growing technologies in 2019.

Companies that are able to gather, analyze and leverage data across their organization will be in the best position to use the technology, and use the real-time insights to drive real-time results—from improving supply-chain operations, to preserving bee hives, to preventing accidents, and beyond.

Learn More: Top Machine Learning Trends – A Year in Review (Western Digital Blog)

This content is produced by WIRED Brand Lab in collaboration with Western Digital Corporation.



  1. Smart Roads: 5 Innovations That Could Soon Become a Reality
  2. Octo Telematics
  3. ApisProtect
  4. Fashion Giant has $4.3 Billion in Unsold Clothes
  5. ClearMetal
  6. Technology, Media and Telecommunications Predictions, 2019


Related Posts