TechSee, which describes itself as an “intelligent visual assistance” company, today announced the launch of Eve Cortex, a platform that teaches itself to recognize thousands of products, models, parts, and components by ingesting only a handful of data points. TechSee claims that by leveraging a combination of AI and synthetic data, Cortex can train itself in a matter of hours, providing end users with step-by-step visual guidance via an augmented reality (AR) overlay.

The AR market is estimated to grow from $10.7 billion in 2019 to $72.7 billion by 2024, according to a recent Markets and Markets report. At least a portion of that growth has been driven by field service applications; technicians are faced with the challenging task of working on equipment with varying technical specifications, often in confined or hard-to-reach spaces. With AR apps, they could have all of the information they need displayed in front of them while keeping their hands free to work.

TechSee was founded in 2014 by Eitan Cohen, Amir Yoffe, and Gabby Sarusi. Cohen conceptualized the idea after struggling to walk his parents through an issue they were having with their cable service. The company’s cross-platform apps employ computer vision to recognize products and issues and streamline warranty registration. Customer agents can see what customers see through their smartphone cameras and visually guide them to resolutions, using either live video or photos.

Cortex builds on TechSee’s existing technologies to enable enterprises to custom-build their own visual self-service flows, without coding. With Cortex, companies can design journeys for product unboxing, billing, contracting, troubleshooting, warranty claims, product registration, technical repair, and more.

“The future is here, and it’s time to say goodbye to user manuals and embrace AR assistants,” Cohen said in a press release. “Virtual assistants that can see, hear, read, and interact are what’s going to transform the way that consumers and employees receive assistance, while saving brands millions and generating new revenue.”

Eve Cortex

Cortex can walk users through the unboxing of various consumer electronics, from security cameras to thermostats, and capture information for upselling while explaining invoices by reading water, gas, and electrical meters. Insurance policyholders can use Cortex to document damage to insured property or identify items they want to insure for virtual underwriting. Moreover, Cortex can certify that an on-site field technician has made a successful repair by examining work through the technician’s smartphone or tablet camera or AR glasses.

Behind the scenes, TechSee employs a combination of few-shot learning, transfer learning, and fine tuning for image segmentation and classification. Few-shot learning refers to the practice of feeding an AI model with a very small amount of training data, while transfer learning is a technique where a model trained on one task is repurposed on a second task.

One of the ways that Cortex learns to recognize products is by ingesting a company’s existing contact center knowledge base. For every device, each article describing visual symptoms and issues, both from customers and field technicians, is extracted and normalized. Then, a computer vision model is trained on synthetic visual data gathered in the lab as well as other visual resources and images supplied by customers, enabling Cortex to analyze, time, and measure the success of each step of every resolution, shortening and optimizing them over time.

To use Cortex, a user taps on a link sent in a text message from a self-service channel and connects via a web browser. They’re guided visually and with voice instructions on how to capture images. After they receive a confirmation message with a reference code, a customer service agent joins, views the images during a live video session, and provides step-by-step on-screen AR guidance.

According to Cohen, companies including Vodafone, Telus, Orange, and Hippo have already tapped Cortex to create new customer experiences. Moreover, tens of thousands of field service technicians in the U.S. are using the platform to install fiber optic boxes.

“One of the very few silver linings to come out of this pandemic is that it’s accelerated adoption for our technology — in the field, at support centers, and for individual consumers,” Cohen told VentureBeat in a previous interview. “Everything is now contactless. When technicians simply aren’t allowed to enter a customer’s home to repair, say, a wireless router, or when a field technician cannot be dispatched to repair an HVAC system, both businesses and customers have to adapt.”