domeet webmaster

“Real Intelligence” held its first product launch conference in July 2020, and officially released “Real RPA” products. At that time, fully self-developed RPA products, especially the process engine, the element picking ability of complex scenes, and the AI-Lab embedded in Z-Factory (process designer), such as the rich AI components and other differentiated innovations, made the majority of developers and end customers present. Yiliang has caused quite a response in the industry. Since July last year, “Octopus·Digital Employees” created through “Real RPA” have been applied in batches in various government and corporate scenarios, continuing to help customers improve quality, reduce costs, and increase efficiency. In the process of delivery and implementation, “Real RPA” products are constantly accumulating and iterating rapidly. In addition, as an artificial intelligence technology company, “Real Intelligence” has rich experience and successful practice in the AI ​​field, and its technological innovation capabilities in the fields of NLP, CV/OCR and intelligent decision-making are in the first echelon in China. In the past six months, it has been continuously advancing the integration of fully self-developed AI capabilities and RPA products, and has carried out upgrades and iterations in terms of technical architecture and innovative applications.

At the press conference, Sun Linjun, founder and CEO of “Real Intelligence” mentioned, “In 2020, the company will complete three consecutive rounds of financing, release four-piece RPA products, and put forward the vision of’contributing 1 million digital employees to the whole society’ , Enter a new fast lane in terms of product technology innovation, market channel expansion, etc. In the past six months, we have accumulated more accumulation and thinking. This time we will release a number of technologies that are not available in the industry at home and abroad. We believe these are based AI’s innovative technology will bring changes to the RPA industry, and we will continue to use high-frequency product iterations to bring the industry’s development to a faster pace, and continue to share new thoughts and practices.”

During the press conference, in terms of “Real RPA” products, “Real Intelligence” proposed the “intelligence equation” to release new nuclear energy for digital employees, E=M×C², that is, through the overall improvement of product functions and experience (M) Superimposed with the power of running speed and design efficiency (C) to realize rapid iterative evolution of RPA products. That is, it has realized the upgrade from a pure AI+RPA technology combination to a product tool for quickly creating digital employees, from a conventional central console to an intelligent hub (process task scheduling control + light data center + intelligent bridge hub), from AI+ RPA is upgraded to AaaS (AI as a Service) + RPA technical architecture. The specific performance includes the “three major upgrades” (all-channel data collection, all-scene trigger mechanism and all-round experience optimization), the “three sharp weapons” (multi-task parallel, SMS verification code and USB Hub), and “Elements Library management, global variable optimization, component annotation function, support for importing Python packages…”, etc., have been iteratively improved in more than 90 items in terms of performance improvement, experience optimization and scene expansion.

Really intelligent “AaaS+RPA” architecture

It is particularly worth mentioning that the AaaS+RPA architecture released by “Real Intelligence” this time can fully adapt to the personalized scenarios of customers for AI empowerment. Digital employees collect authorized data and upload them to the local data warehouse in the process of completing specific tasks. Model training and self-optimization are carried out through the algorithm platform. Then the decision-making platform automatically discovers the trained models and publishes them in one stop, and then directly invokes them by digital employees . In this process of continuous iteration, the model will become more and more accurate, so as to continuously improve the capabilities of digital employees, enhance the adaptability of scenarios, and more effectively deal with the accuracy of general AI capabilitiesInsufficient, unable to use customer data for efficient iterative optimization and other issues.

Another highlight of this conference is the blessing of RPA by AI capabilities. “Real Intelligence” proposed the evolution direction of the combination of AI and RPA as “AI makes the application scenarios and functional experience of RPA fission, that is, RPA’s AI power concept”, and released the following three bright features and four intelligent components (smart document review, credit report identification, medical report identification and VAT invoice verification)

“Beidou”: An easier to use anchorless element picker. Learn the topological relationship between interface elements through graph neural network technology, and then support automatic anchor point selection, so that users do not need to manually specify anchor points when picking up elements, achieving perfect support for scenes such as virtual desktops;

“Aegis”: A non-perceptual process repair technology for digital employees. Automatically find process abnormalities, automatically call the anchorless element picker to complete the process repair, and automatically report the problem. Realize process repair 0 manual intervention, 2 kinds of retry mechanism custom configuration, process repair completed within 10 seconds, effectively coping with 90% of page changes;

“Magic Mirror”: A more efficient intelligent process generation technology. It is different from the current industry that only records the absolute coordinates of the keyboard and mouse operations and elements, which makes it impossible to cope with frequent page changes. Instead, by analyzing the business video, combining with the operation log correction, based on the latest target detection technology and image matching technology, one-click generation of 80% complete business component flow, reducing the process design time by about 30%.

It is reported that in just one and a half years, “Real Intelligence” based on the fully self-developed AI+RPA technology, has served hundreds of large and medium-sized enterprise customers through fast iterative, efficient and stable “octopus·digital employees” products With the government, while winning the recognition and trust of customers, it also helps customers accelerate the process of digital transformation.