Summary of AI Programming Practical Experience
Author: SparkToAI
Published: October 6, 2025
Project Review
In our recent AI programming practice, we have successfully completed four representative projects:
The criteria for selecting these projects were clear: all are within familiar application scenarios, making it easier to quickly validate the effectiveness of AI programming.
Development Process Evolution
Initially, we adopted a straightforward workflow: using DeepSeek to outline requirements and document ideas, then passing the documentation to Qoder for code generation.
During actual development, we found that the interface layout phase often required repeated adjustments, consuming significant time. To address this issue, we incorporated a prototyping design phase into our workflow.
The improved process, which involves submitting both requirement documents and prototypes to Qoder, significantly enhanced development efficiency.
This experience taught us: the clearer the requirements, the better the results with AI programming.
Optimized Workflow
- Requirement Analysis - Use DeepSeek to clarify functional requirements
- Prototype Design - Generate interface prototypes through DeepSeek
- Code Generation - Submit requirements and prototypes to Qoder for project code generation
- Functional Verification - Test complete functionality to ensure it meets expectations
Regarding tool selection, DeepSeek can be replaced with ChatGPT, and Qoder can be substituted with other AI programming tools like Cursor. The key is to find a tool combination that fits your work habits.
Final conclusion: Clear input determines quality output, which is particularly important in AI programming.