Forrester Startup Spotlight 2018 No. 1: AI, Edge, And Privacy
I take dozens of briefings a quarter from vendors trying to snag my attention. Frankly most don’t. In this new blog series, Forrester Startup Spotlight, we will highlight interesting technology startups that do. These up and comers exemplify trends we identified in The Top 10 Technology Trends to Watch: 2018 To 2020. Check out my blog postabout this report to see all the trends.
For this first edition, I bring to your attention three firms that exemplify trends we identified around artificial intelligence (AI), edge computing, and privacy: Anagog, Swim, and Anonos.
Startup Spotlight: Anagog Brings AI To Your Mobile Phone
Trend five from my report is, “Software learns to learn.” In that trend, we stated that the really interesting thing about AI is how it will lead to software capable of learning on its own. Anagog demonstrates how this is happening:
- About the company: Based in Tel Aviv, Israel, and founded in 2010, Anagog has fewer than 50 employees and has raised approximately US$15 million in venture funds. They claim their software is deployed on 25 million mobile phones today and is collecting billions of anonymized data points daily. They have a version of their software development kit (SDK) that is free if you let them own the anonymized data, or you can license it and keep the data yourself.
- Problem they solve: The typical approach to creating insights and predictions from mobile devices today is to pass the data from the device to a cloud or data center repository, do the analysis there, and then round-trip the results to the mobile device. This is very latency-heavy, risks data breaches over the mobile network, and is expensive. Also, the more customers you have, the harder it gets, because you really want a predictive model for each customer. That’s a lot of models. Anagog claims to turn mobile devices into AI-powered digital twins.
- How they can enrich your digital business: Anagog’s technology executes the whole process of predictive analytics on an Android or iOS mobile device in a self-learning, unsupervised way. Moreover, it does this without killing the phone’s battery. Mobile developers can embed Anagog’s offering via its SDK directly into their mobile applications. Thus, the mobile app, in effect, becomes a predictive, digital twin of the phone’s owner. That may seem like a privacy nightmare. But their software only transmits anonymized data back to your servers. They showed me a demo of an Anagog-enabled mobile app predicting a user’s need for a ride share and suggesting it on the spot. No trip to the cloud, no model training required. Their software learns about customers simply based on location and phone use.
Startup Spotlight: Swim Conquers Edge Machine Learning
Trend one in our report is, “IoT shifts computing toward the edge.” Swim’s technology is a great example. It is also another example of trend five, “Software learns to learn.”
- About the company: Swim.ai is a privately held, non-VC-funded firm based in San Jose, Calif., and founded in 2015, with between 11 and 25 employees. Their platform, Swim EDX, purports to “build digital twins that learn on-the-fly, training themselves from real world data so they can predict future behavior for enterprises, equipment manufacturers, smart-cities and IoT businesses.”
- Problem they solve: When end devices don’t have the needed compute power, an emerging pattern is to pass data to an edge server do perform aggregation and some basic analytics, but they tend to be very limited. If you want rich analysis, you still typically have to pass the data back to the traditional data center or the cloud and then build, train, test, and deploy models to the edge. Then you must monitor and govern many models deployed to many servers. The huge variation in edge data and potential analytic insights makes this a daunting task that firms can only afford to do in a few situations. Swim claims to make sophisticated machine learning at the edge easier.
- How they can enrich your digital business: Swim has built an unsupervised machine learning SDK that customers deploy to modestly powered edge infrastructure. The software uses unsupervised machine learning techniques, similar to Anagog, to infer data schema and predict events through correlation analysis. This lets their software work autonomously on edge infrastructure without the need for round-tripping to the cloud. They use the Actor event programming modelto build models, such as of individual traffic light sensors and imagery. Then their software aggregates these to build higher-order models, such as a traffic intersection. Last, their software applies algorithms to automatically optimize model events such as how traffic flows through intersections. It does all this without data scientists or traditional model build-train-test-run processes.
Startup Spotlight: Anonos Solves For Compliance + Analytics
Trend nine in our report is, “Contextual privacy boosts brand value.” Anonos’ solution lets firms build privacy as a first-class citizen of their design; it also helps clients deal with GDPR compliance while retaining the ability to do analytics on customer data.
- About the company: Anonos is a privately held (non-VC funded) startup based in New York and founded in 2012, with fewer than 50 employees. Their product is called BigPrivacy, for which they hold five patents. It purports to “provide users a platform to pseudonymisedata anywhere from the point of ingest to the data lake.”
- Problem they solve: The GDPR requires that firms assert a legal basis for collecting, analyzing, and using customers’ personal data. What is more, once you select a legal basis, you can’t go back and change the basis later. Most compliance offices are focused on the basis of consent; the problem with consentis that you must be very specific. In other words, you can’t ask for consent to explore data and figure out interesting things. This is exactly what data scientists often want to do.Moreover, every time you want to do a different analysis, you have to ask for consent all over again. We think this will get annoying, and users may eventually just opt out. Anonos claims to help customers design for privacy while still retaining the ability to collect and analyze data.
- How they can enrich your digital business: The GDPR is actually conducive to performing analytics, but you have to do it differently. Also, firms will need to use a different legal basis such as legitimate interestto avoid the rat hole of asking for consent over and over. They must also design data collection and analytics such that consumers’ personal data is protected first and not as an afterthought. That’s where Anonos’ technology comes in. They implement a replacement of users’ personal information with tokens in a process called “psuedonymisation,” a term defined by the GDPR in Article 4(5) and used 15 times in it. Firms can analyze and share pseudonymized data for aggregate, anonymous analytics. Or when needed, firms can re-identify it with personal information in a controlled way. We think this product is more than just a GDPR compliance solution, however. It is a product for engineering privacy into your solutions while also being able to analyze data.
New Technology Research Products and Services Coming Soon
To help clients better get a handle on the emerging technology landscape we have formed new team that will provide additional research services focused on the startup ecosystem. For more information on this exciting development, check out Carl Doty’s blog.