5 minutes with Ofri Ben-Porat from Edgify

Can you tell me about your company?

Edgify is developing a unique MLOps platform entirely for edge devices such as MRI machines, connected cars, street lights and self-checkouts - essentially any device that has a CPU or GPU -   - to train deep learning models for a range of industries. The potential of the technology is enormous due to the volume of data each company generates, most of which is rendered redundant due to the processing power required to upload, interpret and analyse it in the Cloud. 

Some self-checkout and weighing machines in grocery stores currently use Edgify’s solution to distinguish barcode-less produce before sharing the acquired knowledge or model across a distributed yet collaborative framework of point-of-sale machines, leading to faster, more accurate and low-touch checkouts. 

What is your role and responsibilities at the company?

As the Co-founder and CEO of Edgify, my overall responsibility is to ensure the success of the company. However, that can sound a lot more glamorous than it is! A big part of my role is to stay in constant contact with the team and our clients, looking at the performance of our technology and the way our platform is used in the real world. I look at lots of data and analytics each day, which helps me to have an oversight on the business as well as the impact of our technology - such as seeing a reduction in theft through false selection at self-checkouts. This is a hugely exciting aspect of my role and reassures me that our technology is having a positive impact on the industry. 

How does your company utilise AI?

Edgify enables businesses to train their AI on the entirety of their data. They no longer need cloud infrastructure or big servers in-store, effectively reducing the risks, costs and time associated with transferring sensitive data to and from an external server. This has allowed industries like retail to train on the entirety of their data, and reach accuracy levels never achieved before, enabling self-checkout machines to achieve 99.98% accuracy versus typical levels of 55-65% through common means of model training. Furthermore, due to the distributed and continuous learning, the rates never decrease. 

Part of the solution we developed is the framework or platform itself for the edge training. To make sure that it can be used by any industry in the future, our own AI team uses our platform to run their research on, and by doing so, improve the capabilities of our framework. 

How can AI and edge devices work together to improve business operations?

By using an edge device for analysis and training of information, rather than the Cloud, reduces the risks, costs and time associated with transferring sensitive data to and from an external server. This allows businesses to train on the entirety of their data, and reach accuracy levels never achieved before.

What can we expect from Edgify in the future?

While Edgify’s technology is agnostic, we have currently been focused on the retail sector. Given the current trend of retail innovation led by giants such as Amazon, the sector is primed for improved processes and new ways of running a store. In time we hope to expand deeper into other industries including healthcare and mobility. 

The final goal for Edgify is to release its Edge MLOps platform for the wider research community. Allowing any company from any industry to run its entire AI stack on the edge.