Jianguo DingSoftware Engineer

Knowing the definitionisn’t the same as understanding.

Here I keep the questions I’ve followed further: put back into concrete situations, compared and tested, and made into something others can check and take further.

Start with the index

Series

Each series follows one question down: from a first intuition to something you can check.

  1. Finance

    From its first P&L to its first M&A, what counter-intuitive problems does a small business run into?

    Experiments
    4 scenario experimentsBusiness decisions in a made-up shop
    Practice
    Cash ReviewA working paper on ICHIBANYA’s real FY2026 filings
    Go to the Finance series in Chinese

    Week 3: profit ¥50,400, cash ¥2,400, receivables ¥63,000.

    Figure · Finance 02 · Scenario 2 · Orders keep growing
  2. AI / RAG

    Why does a RAG system still answer wrong when it found the right documents?

    Experience
    The Paper Room · the basic game
    Topic
    RAG 02 · Chunkingin Chinese
    Go to the RAG series
    Figure · The Paper Room

Projects

Products and systems I’ve built, each starting from a concrete problem.

  • GoodIdea

    For people about to build a product with AI

    Problem
    The idea isn’t clear yet, and AI is already writing the code.
    What I built
    A product definition agent: AI roles for research, design and engineering raise their concerns first, you make the key trade-offs, and the result is a first scope and handoff package a coding agent can start from.
    Today
    The public site has a demo on a fixed example; the workspace isn’t open yet.
  • Forge

    For teams maintaining existing systems

    Problem
    Your coding agent can finish the change. Should it also make the product decisions nobody approved?
    What I built
    Before a coding agent changes an existing codebase, surface what it shouldn’t decide on its own, and let people decide first.
    Today
    A real case on Spring PetClinic and one baseline evaluation are public; the code is on GitHub.