The Recovering AI Skeptic

Daniel Sarosi — founder of Great Wall Connect, advising foreign companies on bringing digital products into China. Born in Germany, Hungarian by descent, Canadian by citizenship, and in Hong Kong for twenty years. Computer science degree with a software engineering specialisation; a JVM specialist across Java, Scala and Kotlin. Previously built an IoT platform connecting close to a thousand textile machines worldwide, running on Kubernetes across clouds including Alibaba Cloud's Shanghai region. Works with the German Chamber of Commerce, and has recently incorporated Chambre Asia, connecting businesses across Asia and Europe.

Chapters

1. 0:00Intro
2. 0:21Back in Causeway Bay — a new studio
3. 1:11Who is today's guest?
4. 2:08Great Wall Connect — cross-border business
5. 3:11Meet Daniel
6. 3:26Born in Germany, Hungarian by descent
7. 4:34Twenty years in Hong Kong
8. 5:04A JVM specialist — Java, Scala, Kotlin
9. 5:27Turning around on AI
10. 6:17What bad code actually is
11. 7:07Good code is code that's easy to change
12. 8:02Why nobody can tell you how long it takes
13. 9:51Starting his own business
14. 10:02Bringing a digital product into China
15. 10:48Octopus, and China's mobile-first leap
16. 11:45The firewall and other hurdles
17. 12:19North and south China behave differently
18. 13:02Building a partner network in China
19. 13:45Hong Kong as super connector
20. 14:22Why he was an AI skeptic
21. 15:05Vibe coding vs software engineering
22. 15:40What 'best practice' actually means
23. 16:26Dreamweaver — when five lines became twenty
24. 17:22AI writing 2,000 lines instead of 20
25. 18:03Data security is the real concern
26. 18:54'They left the database open'
27. 19:34Not locking the door, then posting the address
28. 20:23A sourdough app is fine. A legal database is not.
29. 21:14On Claude Code and the hype cycle
30. 21:57What a responsible manager should do
31. 24:02An IoT platform for German textile machinery
32. 25:34A thousand machines, 40 GB a day
33. 26:02Cutting cost by working remotely
34. 27:27Multi-tenancy — keeping customer data apart
35. 27:54Kubernetes and Alibaba Cloud Shanghai
36. 29:58Healthy skepticism
37. 30:42Talking to Alexander about vibe coding
38. 31:31People have always picked up coding
39. 32:37Vibe coding as a way to learn
40. 33:13Every hype cycle looks like this one
41. 33:33Kubernetes — more servers, fewer people
42. 33:59Claude Code for AWS and the hyperscalers
43. 34:23Letting AI drive the DevOps side
44. 35:12Finding GLM, and using what's local
45. 36:11The Claude Code refugees
46. 36:46Building something, keeping costs low
47. 37:14Ali, the CTO — and why him
48. 38:25What Great Wall Connect actually does
49. 38:41Travelling China with overseas delegations
50. 39:41Solve the customer's pain, don't wait for the call
51. 40:18The factory owner from Guangdong
52. 40:33Not every company has a Tony LC Sign
53. 40:54How LC Sign went viral
54. 41:20“I don’t speak Mandarin”
55. 41:33The 9 a.m. reminder that gets ignored
56. 42:12Stinky tofu, and knowing your foods
57. 44:402008 Beijing Olympics as the dividing line
58. 46:12Two years in Taiwan, then back to China
59. 47:05Not as backward as you thought
60. 47:59Societal rules — Taoism and Confucius
61. 48:43The bus with more passengers than seats
62. 49:46The checkpoint, and organised flexibility
63. 50:19Chambre Asia — the other organisation
64. 50:29Helping Chinese companies find opportunity
65. 51:19Formalising the LA partnership
66. 52:00Automating matchmaking — the concierge service
67. 52:49What Chinese firms actually fear
68. 53:30Compliance, and who do you even talk to
69. 54:07The Gulf, and markets that are larger
70. 54:53Building the portal
71. 55:27Why Chinese websites look the way they do
72. 56:08Aesthetics, East and West
73. 56:35Using AI to rate the websites
74. 57:34AI as a sophisticated pattern matcher
75. 58:30Institutional knowledge and history
76. 58:59“100% AI and secure” — wouldn’t trust it
77. 59:18Thanks and wrap-up

About This Episode

Most of our guests arrive already sold on AI. Daniel Sarosi did not. A JVM specialist with twenty years in Hong Kong, he watched people generate code that broke every practice he had spent a career defending — and said so. This is the conversation about what changed his mind and what did not. Along the way: what bad code actually is in plain language, why he still will not trust non-engineers to vibe code anything that matters, why data security is the part nobody talks about, and what foreign companies get wrong about entering China. Then the second half: what Great Wall Connect does day to day, what living in Taiwan and travelling China taught him about rules and flexibility, and where he actually lands on AI — a sophisticated pattern matcher he would not trust to build a secure system alone.