Instart selected as an Intel Artificial Intelligence Specialist

Instart selected as an Intel Artificial Intelligence Specialist

A fast and secure digital experience is imperative for a website’s success. Fifty-three percent of mobile site visitors abandon a web page if it takes longer than three seconds to load. There are many factors at play when it comes to determining the efficiency and speed of a website, such as image format and resolution, the number of 3rd party services integrated, network speed, geographic location and much more.

Because of the importance of a high-performing website, at Instart, we use artificial intelligence (AI) to help brands optimize their digital experience. Today, I’m proud to announce that as a result of the work we are doing with AI, Instart has been selected by Intel as one of the company’s six Artificial Intelligence Specialists. This is a select group of elite partners that have deep expertise in designing and deploying AI

Solutions on top of Intel hardware. Through its Artificial Intelligence Specialists, Intel is able to offer customers up-to-date insights on the latest technologies and solutions that are powered by Intel hardware with configurations tailored to the customer’s needs for reduced R&D cost.

At Instart, AI allows us to provide brands accurate, continuous insights into browser activity and context-aware control over the web experience at unprecedented speeds. In turn, this helps our customers offer amazing, personalized web experiences to their consumers regardless of the end point they are using or the type of content they are accessing.

Many companies rely on paid technicians that manually manage and troubleshoot issues and identify and correct the factors that are slowing down or threatening the security of their websites. This can take anywhere from several minutes to several hours, and this kind of repetitive processing often leads to tunnel vision and fatigue for developers. In a modern world, web apps are changing on a weekly, if not daily basis. Manual approaches quickly become outdated. As a result, websites slow down and face a higher volume of security threats, leading to internet users abandoning their web pages.

At Instart, we’ve overcome these challenges by using interpretable machine learning to take a task that a human technician would do slowly and ramp it up at a massive scale. For example, Instart uses AI to analyze the visual quality of images and intelligently determines if and how to compress them to optimize load speed without compromising the quality to the human eye. This results in a website that loads faster but maintains integrity.

We use expert human technicians to train our AI. In the case of our image optimization system, we took a catalog of over 2000 images and had a panel of experts train the AI on the best image settings to use for each. The system then discovered 35 unique types of images found on the web, such as product images with complex patterns on a white background, humans wearing clothes and jewelry at medium distance in the scene, or far-away shots of buildings. After the system was trained it could then analyze new images, match them to one of the 35 categories, and then select the optimal image settings based on the actual content in the image in seconds.

Once this process is complete, the AI is able to review billions of images or security events in seconds, significantly decreasing the analysis period. Our system can even explain how it came up with a solution,  revealing the image category and image settings selected.

As you can expect, the early versions of this system, which we call SmartVision, were very processing intensive. We determined early on that to get from a proof of concept in the lab to a high scale production AI-powered service we needed to optimize the computer vision, AI, and even the image transcoding systems. We undertook an extensive evaluation of hardware options, including high-end GPUs and expanded hardware acceleration capabilities from Intel. We found that the Intel Xeon family of processors provided the best combination of performance, efficient power consumption, and compatibility with the existing server and software solutions used in our architecture. As a result, we have standardized on Intel Xeon as our platform of choice for our global cloud service.

Our existing experience leverages Intel Xeon technology, and we’re excited about this new opportunity to work with Intel and its customers to revolutionize the digital experience. By selecting Instart as one of its Artificial Intelligence Specialists, Intel is offering its customers the opportunity to build a fast and safe web experience optimized by cutting-edge AI technology.