This post continues from my previous LinkedIn article: https://www.linkedin.com/pulse/reflections-6-levels-ai-assisted-coding-h%E1%BA%A3i-nguy%E1%BB%85n-ng%E1%BB%8Dc-cdvbc/?trackingId=ekSgqVRiCjOGsBJR%2B41MXA%3D%3D
After spending a long time using AI to build products, I’ve managed to complete many things that are extremely challenging from a technical perspective. What I’ve done is honestly not easy to summarize in a short and clear way. I want to share it, but if I talk about the products or the implementation details, it becomes too long and not easy for many people to follow.
So I tried something different: I asked my AI Agent to assess my skills based on what it has observed from our work together. And I was genuinely surprised by what it said. I’m sharing it here because it made me happy and also because I hope it may help others reflect on their own journey.
Assessment PDF: https://weebpal.com/hai-nguyen-ai-development-assessment.pdf
Why I’m sharing this
My purpose is to look for opportunities to collaborate with people who truly have strong business experience and vision, people who want to develop product lines together.
On the product-building side, I’m confident about what I and WeebPal have been doing. Building and deploying software systems is not a big issue for us. What I’m really looking for is help, guidance, and partnership from people with strong business capability, so we can build something meaningful together, faster, and in the right direction.
A quick summary of the long path behind my AI work (and related experience)
I want to add one small context before the timeline below: what I can do with AI today is not the result of a short-term push. It has been a long journey, and I feel lucky that each step gave me a piece of foundation that gradually came together and helped me collaborate with AI more effectively.
- 2001: I started learning programming. I joined many algorithm and programming competitions, and I was lucky to win some national-level awards.
- 2008: I did my university thesis on Speech Recognition.
- 2013: I did my master’s thesis on controlling a computer by voice, and I built a working system myself.
- 2012: I founded WeebPal with a product + customization direction, starting from zero: no clients, no money, no market, no product… I spent nearly two years working overtime to build free products for the community, then gradually moved to premium products to find real customer needs.
- 2014: I started working full-time with WeebPal Technology Corporation. After more than 12 years, we have completed 200+ projects: from basic websites to government-level projects (UK, Australia, Switzerland, France, the U.S., the Samoa government, ...), and enterprise work (Petronas), etc. However, most of these projects were delivered through outsourcing under partner names, so in many cases, I still remain a “hidden outsourcing supplier.”
- 2019: I had the idea to build an AI-driven system with sibu.ai. Unfortunately, I didn’t have enough resources to bring it fully to market. But it still gave me a complete end-to-end “lab-level” system for building an automated website pipeline. I wrote this blog in December 2019 (long before OpenAI became widely known): https://weebpal.com/blog/how-does-robot-build-website
- 2022: When OpenAI became very popular, I tried it, but at that time I didn’t find it effective for development, so I didn’t use it deeply in coding. However, I used it heavily as an analysis tool, and it helped me a lot in project design and architecture thinking.
- 2025: I returned to AI again, and I was truly impressed. At this point, AI could solve almost everything I needed. I started with simple problems, then increased the difficulty many times. The results made me extremely satisfied. I wrote this blog to share my full perspective on AI capability: https://www.linkedin.com/pulse/reflections-6-levels-ai-assisted-coding-h%E1%BA%A3i-nguy%E1%BB%85n-ng%E1%BB%8Dc-cdvbc/?trackingId=ekSgqVRiCjOGsBJR%2B41MXA%3D%3D
- Now (March 1, 2026): I’m using AI to control projects fully at Level 5. Combined with my experience in analysis and system design, I can manage the complete process of turning a business idea into a complete system with close to zero manual coding, in a very short time.
In my view, using AI is not about asking “what can AI do.” AI can handle a surprising amount of what developers do, often faster, when the problem is framed clearly and the system is designed properly.
The most important part of using AI can be summarized into three points:
1) You must know what you want and describe it clearly You need to express what you want in a clear, logical, and consistent way. You must also be able to review the output, accept it, and request improvements until it becomes exactly what you want.
But this is still not enough. AI Agents always operate under constraints (context, tooling, compute, time). So if you throw a very large problem into it in one shot, what you receive can be a big disappointment.
2) You must be able to break the problem down within a consistent architecture You need the ability to analyze the big thing you want into smaller problems, but still inside the overall logic and architecture. And as projects grow, this becomes much more important: you also need to truly understand data structure, and optimize data structure. Only when you master and optimize the data structure can you understand how the core system works. Then adding/removing features and scaling requirements becomes something you can control.
3) You must think like the user (this decides whether a system survives) This is the deciding step for a new generation software system to be useful and to survive. In the past, software was created manually, and many limitations were actually limitations of resources and cost. With AI, that limitation is solved very strongly—so “finishing what is not finished yet” is not hard from a cost perspective. But from the builder’s thinking perspective, it requires foundation, real experience, and creativity at a high level.
I’ll end with one sentence I used in the previous blog:
AI does not replace thinking and architecture, AI amplifies the value delivered by thinking and architecture.
And finally, here is the main quote from my AI Agent about how anyone can create a similar assessment:
“Anyone can create a similar assessment by pair-programming with AI on their own real projects, then asking AI to evaluate based on what it has observed.”
If you have experience in business strategy, product direction, go-to-market, partnerships, or scaling, any guidance or connection would mean a lot to me. I’m still learning on the business side, and I’m actively looking for the right people to learn from and build with.