In the past few years, the market for artificial intelligence (AI) and machine learning technologies has gained strong momentum. What's interesting, though, is that the much of the innovation in this space is driven by disruptors, not legacy vendors.
These 'upstarts' are companies born in the internet age or companies that have transitioned into the AI market, and are building out useful AI products that will likely broaden in their impact over time. Much like how Amazon Web Services (AWS) became the infrastructure provider of choice for many companies with the rise of the public cloud, many of these upstarts will see their products widen in application.
While many of these companies might have originally created their product for a specific use case, it's possible that they will grow into platforms on which the company may build an additional revenue stream. Here are five upstart companies that are pioneering the use of AI and machine learning in their businesses.
It's no secret that Uber considers itself more of a software and technology company than one that merely schedules rides. A big part of that identity is a heavy reliance on machine learning and artificial intelligence. In fact, Uber's head of machine learning, Danny Lange, spoke at the Applied Artificial Intelligence Conference this summer.
The amount of data produced by a company like Uber is huge, and machine learning helps executives to leverage that data for business value. Uber isn't extremely open about how it uses the technology, but it likely helps with their logistics and many other aspects of the business.
"They can mine all that data to figure out how to retain their customers, how to increase the value that the customers are getting, how to prevent churn," said Kalyan Veeramachaneni, principal research scientist at MIT's Institute for Data, Systems and Society.
Additionally, as 451 Research analyst Nick Patience noted, Uber's use of machine learning will likely expand as it continues to invest in autonomous vehicles.
"That's because autonomous vehicles require the ability to be constantly learning about their environment, be it the roads, people, other vehicles, the weather, etcetera," Patience said. "This would be impossible to achieve if this all had to be hand-coded into the application; it can only be done with the applications learning about their environment themselves and adapting to it."
Much of this was detailed in a September blog post from Tesla, which explained how the company is now using radar as a primary control sensor, with cameras and additional technology as supplements. However, AI comes into play in helping the car more clearly understand the objects in its way.
As noted, the company collects data from the whole fleet of Tesla vehicles and uses it to improve the Autopilot experience. In the post, Tesla refers to this as 'fleet learning'. This is especially useful in determining which perceived objects require braking, and which do not. The car collects data on how non-Autopilot Teslas are behaving in a certain area, and weighs that against the action it would have normally taken.
"The car computer will then silently compare when it would have braked to the driver action and upload that to the Tesla database," the post said. "If several cars drive safely past a given radar object, whether Autopilot is turned on or off, then that object is added to the geocoded whitelist."
Salesforce is one of the largest tech companies on this list, but it does deserve recognition. In September 2016, Salesforce introduced its Einstein platform, which is customizable for customers and uses "machine learning, deep learning, predictive analytics, natural language processing and smart data discovery," according to a press release.
The introduction of Einstein highlights a larger trend among tech providers who are using AI and machine learning to improve their existing products and services. However, the company is also using the technology to improve its own business.
"Salesforce is making use of customer data including emails within Salesforce, activity data from tools such as Chatter, as well as external sources like social media and signals from IoT devices to train its machine-learning models, which can in turn drive features within applications, such as predicting churn, predicting close rates and real-time personalized marketing," said 451 Research's Patience.
On the hardware side of things, Nvidia is pushing hard to be the provider of choice for AI solutions in the future. The company is no stranger to new tech trends, but it has recently made a strong pivot to invest heavily in chips made specifically for AI and machine learning applications.
To coincide with the rise of graphics processing units (GPUs) used in AI, Nvidia revealed its Tesla P100 GPU back in April, which reportedly cost more than $2 billion to develop. On stage at the announcement, Nvidia CEO CEO Jen-Hsun Huang said, "Deep Learning isn't just a field or an app. It's way bigger than that. So, our company has gone all in for it."
A few months later, the company followed up the P100 GPU with two more chips. The Tesla P4 and Tesla P40 both made their debut in September.
While Ayasdi leans more toward the startup side of the spectrum, the company is making serious headway in developing next-gen AI platforms for the enterprise. According to Patience, the company has shifted from offering a toolkit of sorts to providing real business apps driven by machine learning.
"It now focuses on applications for finance and healthcare, for example Denials Management and Avoidance, a healthcare application designed for surfacing patterns in denial claims by analyzing attributes," Patience said. "Or, Regime Forecasting and Signal Creation app, which is designed to reveal the precise drivers of returns, risk and liquidity in order to create better predictive models."