Our History

May 28, 2018
Pioneering deep learning

Pioneering deep learning

Allegro.ai’s vision is to empower every company to leverage the power of AI everywhere.

We focus our efforts on computer vision, powered by deep learning. Our goal is to provide a comprehensive platform and intuitive products which our customers can leverage to create products for their users with human-like awareness of their surrounding environments.

We look to empower our customers to develop, deploy, service, and maintain cutting edge computer vision based products that are best of class, continuously kept up to date and specifically optimized per device / user.

Allegro.ai is the brainchild of Moses Guttman, Gil Westrich and Nir Bar-Lev. In 2014 deep learning, an experimental branch of AI, began to show promise for revolutionary application in the computer vision space. As the field rapidly began to expand, the visionary trio identified three fundamental realities:

AI, and DL specifically, is an inherently different software product paradigm. As such, unlike the traditional S/W space, DL missed an entire set of tools to support scale, quality assurance and economics throughout the product lifecycle. Tools / platforms / infrastructure which are imperative if companies intend to roll out deep learning-based products at scale.

While artificial intelligence is a learning-based paradigm, everyone working on AI was essentially making learned (static) solutions as opposed to learning (dynamic) solutions. The learning always stopped when products were deployed from the lab into the field. Why not enable companies to extend the learning throughout the life of the product, especially after it has been deployed?
Real-time computer vision solutions often require the processing / decision making to happen at the edge. Deep learning is a very computationally heavy algorithm. To be able to support personalized models per device, tailored to its specific environment, is the only architectural solution that spans device types, processor vendors and processor version. As a s/w solution to the problem DL therefore carries the highest ROI and widest use-case applicability of potential alternatives.

With these realizations, the three teamed up in 2016 to embark on a journey to address these challenges and opportunities with a vision of enabling every organization to leverage the full power of deep learning to build solutions for a better world.
The three founders raised their initial investment led by Bosch and Samsung in early 2017. In early 2018 the company closed its Series A funding led by Mizmaa Ventures which included participation of all of its existing investors.

From the get-go, the founders realized that a best of breed platform requires a combination of deep domain-expertise as well as key software engineering in several areas. With this in mind,the partners set out to build a 360-degree team surrounding the problem statement – bringing in key talent in deep learning, high performance computing, backend, devops and embedded development.

During its short history, the company has already been able to demonstrate, working video comprehension solutions running on edge devices. In these PoCs, the company demonstrated not only object tracking and detection, but situational analysis, in real time. These solutions were made possible due to the robust internal platform which the company developed.

During 2017, the company started placing its initial platform offering at a handful of customers, allowing them to leverage the first end-to-end platform for AI development & production, starting with computer-vision: automatic annotation, dataset lifecycle management, platform agnostic neural network optimization, distributed training-task automation, versioning and collaboration tools as well as post-production continuous learning & personalization. All provided either as a SaaS solution or on-premises.

Going forward, allegro.ai continues to reinforce the platform, packaging a variety of products for a AI-DL-CV solutions for an increasing client base as well as continuously implementing features for optimization throughout the development-production-deployment and post-production life-cycle.

Scroll to Top