AI

One diagram, and the box at the bottom of it I spent a while putting this together. It answers a simple question: when I say “get this done for me,” which protocols does that sentence pass through before it becomes something that actually happens in the world? Six layers, top to bottom. It starts at human intent and lands on robot arms, sensors, and PLCs. Every layer in between has a name, a responsibility, and a spec taking shape. ...

The exam is done. Last cycle ended with “next up is finishing the Intermediate subject content, plus the AIoT certification I registered for on a whim — after that, back to normal.” These two weeks were basically that list, worked through in order. The rhythm was nothing like last cycle. Then, 96 commits landed almost entirely in one repo; exam prep had squeezed everything else out. This time the commit count dropped by 40%, but spread across four projects: finishing the question banks, two new proposals, a format upgrade in the knowledge base, and the personal site. Going from “only one thing” back to “several things” is itself the signal that prep is over. ...

The first week back from Canada, I attended two conferences — though jet lag pretty much had its way with me for nearly the whole week. Things only started improving this week, and right on its heels came prep for the iPAS AI Application Planner (Associate) exam. I’d signed up around the tail end of AIA’s AIPM course. I happened to see information about the certification on Facebook, and while the iron was hot I registered for both the Associate and Intermediate levels at once — and roped in some AIPM classmates too. So before leaving the country, on top of wrapping up the client project I already had, I also stood up a quiz and drill site using RAG and shared it with the group. Then came a full 7-week cooldown, and it wasn’t until this Monday that I actually started drilling for myself. ...

Over the past two weeks (June 1 to June 14), several product lines have progressed on GitHub. Looking back, the theme is very clear: I’m not just writing features, but organizing the messy information and workflows of real life into systems people can use with peace of mind. MatrixTradingSystem (MTS) Trading Operating System A lot of time these past two weeks was dedicated to the final sprint of MTS, focusing primarily on “preventing real users from stubbing their toes.” Aside from a few last-minute client features, we also added support for the emerging stock market (興櫃). This turned out to be quite interesting — I hadn’t realized that broker support for emerging stocks varied so wildly, which led to some rather peculiar issues. Most of these have now been resolved, with only Yuanta Securities remaining to clarify final details via email (to the point of having to reverse engineer the broker’s App to locate the root cause). ...

When you run a company alone, with an army of AI Agents at your side, the biggest risk isn’t actually “moving too slowly.” The biggest risk is turning yourself into a faster task machine. Over the past half year I’ve been pushing the trading system, Lectio, AI Email, the Mira assistant, several small tools, and knowledge bases all at once — and AI really has exploded my output. But one thing keeps getting clearer: speed itself is not the goal. So I wrote myself an operating directive to remember what AI is actually for. This post is its essence. ...

I used AI to pull together a retrospective from GitHub, and it turned out there was more progress than I’d imagined. Looking back on these two weeks, the point isn’t any single project but several threads moving forward at once. MatrixTradingSystem This stretch was mostly about pushing the trading system toward “something operators can genuinely trust.” A good number of features were added, and it’s now entering the wrap-up phase. It’s not just about features running — the flow has to be smoother, risk controls clearer, error messages understandable, and the risk of misclicks and misjudgments during operation reduced. Next up is shipping this one out. ...

Over the past few weeks, I shipped a new macOS tool called Lectio. The starting point is simple: when you are reading a document, course, paper, product guide, or error message, the moment of confusion is usually small and immediate. Maybe one sentence does not make sense. Maybe a term is unclear. Maybe a stack trace needs one more explanation. Maybe a paragraph depends on context you have not fully connected yet. ...

MatrixTradingSystem Pushed the system from just runnable to actually testable for users. Fixed many real-world issues: broker integrations, order state synchronization, Windows installers, macOS releases, error diagnostics, log bundles, and various preflight checks. You really can’t see these things in a demo; they blow up while testing, and you fix them on the fly. I can only say the Windows environment is truly magical, and while some of these financial brokers are quite modern, others are very old-school. I hope the flag I set can be reached. For the next two weeks, my main focus will be getting this project done and shipping it! ...

It’s time for another bi-weekly review. With AI compressing the scale of time so extremely that even a single day can feel long, why do I still review on a two-week cadence? I thought about it, and there’s no real reason — this scale just feels comfortable. These two weeks, I kept meeting up with friends (and freeloaded quite a few meals, ha), attended a few seminars, continued advancing through the AI PM course and some online classes, reported back to my parents, and went through the move-in process for the studio. Looking ahead to May, a batch of hardware SKUs I’d ordered earlier will start arriving. Most importantly, there’s a small toy robot — I think it’ll be a fun theme, and something my daughter can play with as field testing. Another important thing is building a “real” Agent that can help me run development work in parallel, actually scaling a version of myself. And finally, an old classmate asked me to build a multi-broker trading system. I do invest, but building an order-placing tool was something I’d never considered before; after diving in this week, I found it quite interesting — and I have to plant a flag to wrap this project up in May! ...

AI’s Center of Gravity is Wrong Open any tech news feed and it’s all cloud AI, all the time. Bigger GPU clusters, more training data, pricier API calls. As if intelligence can only live in faraway data centers, available on demand — as long as your network holds up. But that’s not how the real world works. A factory camera needs to spot defects in milliseconds. A vehicle sensor can’t wait for a cloud round-trip before deciding to brake. A remote agricultural monitoring station might not even have a stable 4G signal. These scenarios — factories, cameras, sensors, vehicles — are the backbone of the physical world. And they all run at the edge. ...