What to Decide Before the Next Project

I do not want to be only a researcher or an app developer. I want to define a technological direction and build a company that creates a new market around the technology and products it starts. The end goal is a platform that many people use, where revenue and competitive advantage compound over time.

That means the decision in front of me is not a list of projects. It is the future I am willing to bet on for the long term. Projects can change quickly; the direction in which technology, users, and an ecosystem accumulate should not change casually.

The First Bet I Want to Test

The direction I most want to test is next-generation computing × human-computer interaction. Computing itself may change in form and execution, while the way people perceive information, make decisions, and act through computers may change with it.

This direction leaves room to go deep into hardware, embedded systems, runtimes, AI, and new interfaces. It may also offer a path to begin with a killer product that consumers can experience directly. This is not a final commitment. It is the hypothesis I want to try to disprove first.

There are related directions worth exploring as well:

• Next-generation computing: semiconductors, computer architecture, operating systems, compilers, and new devices
• Human augmentation and new interfaces: smart glasses, real-time multimodal experiences, spatial computing, and new UX
• Physical AI and autonomous systems: robotics, embedded systems, real-time control, and autonomous agents
• Energy and physical infrastructure: power electronics, energy storage, grids, and industrial control
• Self-evolving intelligence: continual learning, world models, autonomous learning, and verification-learning systems

Self-evolving intelligence could be an independent bet, but it could also become core technology that supports several of the directions above.

Technology Alone Does Not Compound

Starting early is not enough. I need an advantage that grows stronger over time.

Proprietary technology can create differences in performance, cost, reliability, or new capabilities. But it must connect to real customers, repeated use, revenue, and feedback. Shared layers such as SDKs, APIs, hardware, runtimes, and IP can then accumulate. An ecosystem moat begins when developer dependence, network effects, and switching costs emerge.

I want to keep asking three questions.

1. Will this advantage remain after GPT-8? More than code generation itself, I need assets that endure: customer relationships, real performance, physical technology, distribution, operations, and ecosystems.
2. Can this compete with big tech? Instead of confronting incumbents with the same resources, I should enter through an asymmetric wedge in a small market, build a distinct advantage, and expand from there.
3. Do the technology and product reinforce each other? New technology can create a killer product, and product success can fund and support research on the core technology.

Moving from a Thesis to a Company

In the first 30 days, I will explore two or three candidates deeply: their technology, market, competitive structure, experts, and future changes. The goal is to state not only what I believe, but also what evidence would disprove that belief.

In the following 30 days, I will anchor in one industry, experiment and build directly, and investigate users and the market. The point is to learn whether I can genuinely immerse myself in the field and whether a product opportunity exists.

Between days 60 and 90, I will release a first product in the form that fits best: an app, developer tool, hardware product, or B2B product. The goal is to test whether people have a reason to use it and pay for it.

Over the long term, I will secure runway through revenue and investment, improve the core technology, and bring in the expertise that is needed. When there is an opportunity to expand from a product into a platform, I will invest in a structure where technology, users, revenue, and ecosystem reinforce one another.

The key is neither to wait until I become an expert before founding a company nor to build products without expertise. It is to accumulate expertise and unique technology while building products.

What to Let Go of Strategically

Rather than starting projects for every fashionable technology, I want projects that build assets under one thesis. Rather than treating implementation difficulty as a moat, I want an advantage competitors cannot easily reproduce. I do not want to keep research with no customer value or products that have no connection to the long-term goal.

Not everything should be discarded at the same speed. Projects and technical approaches can be stopped or changed quickly. Markets should be adjusted when the evidence demands it. A long-term thesis, however, should change only after enough disconfirming evidence accumulates.

The criterion I want to keep is one sentence:

Bet on a future that will matter in 10 to 20 years, survive by making what people want today, and compound the technical and business advantages required for the next stage.

If a project does not build toward that, I should be able to let it go even when it is enjoyable or appears to be working in the short term.