Focus on next-generation software and business models

About Us

We are a team that transforms from software development to software operation

The software we developed directly serves 100,000+ users

The following are part of the software we designed, developed and operated by ourselves, excluding development cases for customers


Ai topic

5G + Industrial Ai image detection

Based on 16 megapixel industrial grade cameras for detection, analysis and artefact evaluation of production line defects without material. The architecture layer is put into production using a combination of 5G + edge computing. ai level uses new convolution to cope with irregularities in defects such as deformable convolution, uses FPN to improve detection of small targets and introduces Transform to enhance detection capabilities.

Indoor mobile risk model

Using computational modeling and algorithms to help urban interior design and planning is a new development direction and demand in the design industry. In order to help make unintuitive and invisible algorithms intuitive and understandable, we have developed visualization tools for indoor crowds and indoor maps. We use the Guangzhou Airport as the base, apply particle swarm and neural network-based model simulations to it, predict the flow of people in various locations, and take rest, security, and boarding into account; and use this to assess the risk of crowd trampling. And the risk and loss of infection.

Big Data Model for Intelligent Transport

To aid the science of urban transport planning, we have developed a customised urban simulator especially for municipalities. The simulator contains global map data as well as simulated vehicle data scenarios within Beijing. We have solved the problem of loading and transferring data in big data scenarios using slice reading/loading and remote procedure call loading, and have controlled the detailed information of each small point while processing the large amount of data, taking the individual vivid individuals and merging them into a meaningful ocean of data.

Industry order OCR recognition

There is a large amount of non-standard order data in the traditional industry procurement supply chain. Through the process of angle proofing, OCR positioning and confidence algorithm, automatic table formation, field matching and data cleansing, the business layer can obtain a fresh and standardised data structure and adapt the data level to the business according to the industry configuration rules (than high temperature lamination, electrophoresis and other processes).

Partner Company






Future Bridge




Tsinghua Tongfang






Peking University

NEAR Tinker Union

Enleap Launchpad

Waver Finance