AI Surfing Tips - A Product Developer's Life with AI

「We shape our tools, and thereafter our tools will shape us.」

In 「AI Surfing Tips」, I dive into my experience as a developer navigating the wild waves of AI. From generating code to streamlining workflows with tools like Cursor IDE, I explore how AI boosts productivity, the hurdles of mastering prompt engineering, and the importance of preserving code quality. I’ll share insights on what tasks to hand off to AI, what to hold onto, and the new skills we need to thrive in this ever-changing digital sea, all while harnessing futuristic tools like ChatGPT and Windsurf to enhance our craft. As with every tech revolution—where many skills become obsolete—I’ll also reflect on how we can survive as individuals by adapting, learning, and staying ahead of the tide.

1. Daniel Teng, 04/14/2025
AI Surfing Tips
A Product Developer’s Life with AI

2. •Work
•セレンディピティ/ Serendipity K. K.
•Wowsei Dynamics Inc
•Home
•Los Angeles & Tokyo 2024
•Tokyo 2022
•Hong Kong 2015
•Qingdao/Jinan/Shanghai
Daniel Teng
ダニエル タン

3. CSPO 2025/04
•敦南 2014/04
•22 CSM courses
•14 CSPO courses
•4 BTS courses
•Students
•569 CSMs
•294 CSPOs
BTS 2024/04
11 Years in Taiwan
BTS 2023/10
CSM 2023/04
BTS 2022/10
CSM 2022/10
CSM 2022/11
CSPO 2022/07
CSPO 2019/04
CSPO 2018/12
BTS 2018/08
CSM 2018/05
CSM 2017/12
CSM 2017/08
CSM 2017/05
CSPO 2017/05
CSPO 2017/04
CSPO 2014/04
CSM 2014/04/17

4. Experience of AI

5. Phase 1 Another Useful Tool

6. Code Gen
•Json -> C# Class
•Copy & Paste from ChatGPT

7. Phase 2 Junior Developer

8. Autopilot
•Integrated IDE
•Better Intellisense / more code
completion
•Small algorithm, functions
•Limited context
•Unsubscribed after 2 months

9. Phase 3 Partner

10. Pairing
•Cursor
•Subscribed -> Unsubscribed
-> (Re-)Subscribed
•Windsurf
•Demo

11. Roles of AI in Life
•Code Gen / Code Review
•Junior developer who take care of small functions
•Junior tester who take care of test cases and test data
•Unfamiliar with grammar and syntax
•Regular expression
•String, date manipulation
•Multi-thread
•Tools commands
•Shell, Cloudflared, PostgreSQL, Docker…
•Unfamiliar languages, libraries, tech
•Php
•Front end
•Trouble shooter
•Design and architecture review, discussion
•Refactoring
•Document

12. Benefits
•Productivity
•Tab-tab-tab
•Shorter cycle from idea to
implementation
•Integrated knowledge & work space
•@File, @Web, @Doc, @Git history…
•More control, less dependencies
•More focused

13. Challenges
•Lack of visibility / control /
understanding over AI-generated code
•Maintenance & Iteration
•Hallucination
•Prompting
•Too big, abstract
•Un-structured
•名词奴隶 Slave of Terms

14. AI Survival Tips

15. English Skill

16. Articulating Clearly

17. Thinking Clearly

18. Programming Skill
•Don’t rely solely on AI to do the
work
•Logic thinking / design
•Validation
•Review / Diff
•Debugging

19. Engineering
•打穿力 BTS (Break Through Starter):
•Idea / Requirement -> Planning -> Design ->
Implementation -> Test -> Operation
•DevOps
•Architecture and module decomposition
•Data structure, mapping, flow
•Tool chains and automation
•Refactoring and evolutionary design
•Baseline / Version control
•Snapshot of progress of work
•Diff debugging
•Compliance

20. Patterns and Best Practices
•Domain patterns
•Architecture patterns
•Integration patterns
•Deployment patterns
•Design patterns
•Implementation patterns
•Clean Code
•Kata patterns
•Refactoring

21. The Very Few Professional Expert
•Go deep
•Business / Industry domain
•Resources
•Tacit knowledge
•Taste
•Go broad - 打穿
•Understanding of business, product, goal,
problem, operation, customer services
•Iterative and incremental SMART Goals
•Slicing of goal and problems
•Mastery of tools of various roles

22. Three Amigos 三剑客
Unit of AI Enabled Organizations
•Business/Product - Captain
•Business sets direction and makes
decisions.
•Developer/Operator - Navigator
•Developer/Operator makes it real — and
keeps it running safely and efficiently.
•Designer (UX/Service) - Shipwright
•Designer ensures the product is human-
friendly and intuitive — increasingly vital
when AI can confuse users.

23. AI Working Style
Cycle

24. AHA Cycle
An iterative development
approach that uses trace logs
as an approval mechanism,
leveraging AI and baselines to
ensure controlled, verifiable,
and incremental improvements.

25. Assess
- AHA Cycle
•Prompt Cursor to add
descriptive trace logs to current
system
•Feed the current system state
(trace log) as a baseline in
Cursor

26. Highlight
- AHA Cycle
•Prompt Cursor to propose
changes based on logged state
and system goals
•Review and compare (diff) the
changes proposed by Cursor
•Accept / Reject / Revise the
modifications

27. Approve
- AHA Cycle
•Run the system and compare new trace
logs with baseline state
•If the system doesn’t work as expected
•Feed logs to Cursor for
troubleshooting, then return to step H
•If the system works as expected
•Feed logs to Cursor for a final review
and refinement & Refactoring
•Ask Cursor to generate documentation
and tests for maintenance

28. AI Surfing Tips
•English Skills
•Articulating Clearly
•Thinking Clearly
•Programming Skills
•Patterns and Best Practices
•Engineering Skills
•打穿力
•Professional Expert
•Three Amigo
•AHA Cycle
•Assess -> Highlight -> Approve