Junxian Ma / Writings

Summer has just begun, and I spent some time thinking seriously about what I should do in the next stage. Over the past period, I have been wavering between research and market. On one hand, I really want to get closer to the most frontier technical problems and understand what people at the frontier are actually thinking about. On the other hand, I do not want to stay only inside technology and lose my sensitivity to real problems, real needs, and real-world feedback.

So this summer, I think it may be a better path not to rush into labeling myself as someone doing “research” or someone doing a “startup.” What matters more is putting myself into a strong enough information field and close enough to real problems, so that I can train my own judgment.

First, I hope to work with frontier researchers for a period of time. Many truly important judgments require entering the actual research process: discussing problems with strong people, breaking down directions together, and running experiments. Only through this can I better sense what is genuinely frontier and what is merely surface-level excitement.

Second, I hope to keep my sensitivity to the market. Even if this stage leans more toward research, I do not want to sink completely into technology. Each week, I want to continue talking with three or four entrepreneurs, investors, and builders whom I find impressive, so that I can understand what problems are happening in the real world, which needs exist over the long term, and what new product forms may be opened up by technological changes.

Third, I hope to gradually form my own AI thesis. On Agentic AI, reasoning models, and next-generation intelligent systems, I cannot simply follow other people’s judgments. I need to continuously record and update my own views: where future opportunities may be, what the current bottlenecks are, which directions are worth long-term commitment, and which problems may be truly important.

Fourth, I hope to complete at least one small feedback loop. Instead of trying to build an ambitious product from the very beginning, I want to choose a concrete problem, build a prototype, let real users try it, collect feedback, and iterate. This process may help me understand the connection between research and market better than thinking alone. Being able to see feedback signals feels very important to me.

Ultimately, I hope to do something big. But the more ambitious the goal is, the less I should remain only inside grand wishes. What truly matters is to keep moving closer to the frontier, keep touching real problems, and keep revising my own judgment.