Development service · Acquisition
Embedded AI Service
The story
The author's team was developing an AI service on an rk3576 device. The client maintained its own mobile app and cloud service. The app connected to both the device and cloud, but the team lacked the cloud communication protocol, making simulated testing difficult.
The author used REA to analyze the client's Android app and recover protocol information. An agent then used those details to write and complete an end-to-end test program. Fees and hours saved were undisclosed.
Source excerpt
今天体验了一下REA： 我们在给客户开发一个运行在 rk3576 的 AI服务，客户自己开发手机app和云端…
@xingshizhai · Read the full source
- What was delivered
Client AI functionality running on an rk3576 device.
- Customers
An embedded-AI client maintaining its own mobile app and cloud service.
Lessons and acquisition
Analyze an app to fill missing integration documentation, then generate an end-to-end test program.
Costs, risks and open questions
Protocol coverage, test scope and actual time savings are unknown.
Tools and topics
- REA
- rk3576
- Android
When citing this case, credit AI Income Journal and include a link to this page. Retain the original author's credit and source link. Content use and attribution