The site caught up with the work

ProjectWong had fallen behind its own projects.

Compass still read like an active job search. PersonalOS still described the earlier second-brain concept. My bio framed this as a stretch of consulting and independent building, even though I had accepted a new role.

Each description had been reasonable when I wrote it. Together, they were telling an outdated story.

Tonight I brought the site forward.

The biggest update was PersonalOS v2. Its new build-log entry explains why I’m giving AI more authority, how I want to use the system throughout my day, and why performance has become part of the rebuild. It also separates the cross-device productivity I already have from the more connected daily experience I’m still building toward.

Compass needed a different kind of update: an ending. The job search is complete, its relationship-management concept is becoming part of PersonalOS, and the Compass name now refers to the MCP and vault access layer. I published a closing entry and marked the standalone project completed.

That decision needed to show up beyond the article. Compass moved out of the homepage’s active-project cards. Its description changed. The bottom of its project page now points readers toward PersonalOS instead of promising more entries as the build progresses.

The older posts remain intact. I added dated notes linking them to the newer chapters, so someone landing on an earlier article can understand what changed without losing the original account of what I believed at the time.

The homepage and experience page needed the same treatment. I’ve accepted a new role, and the independent work continues alongside that next chapter. The site now says that.

I also removed the homepage counters.

They had shown lines of code, documents, and tokens processed. When I added them, I saw those numbers as evidence of how much I was building with AI. Billions of tokens felt like something worth putting on the page.

I’ve come to believe that burning tokens is no longer a point of pride. In the process of burning billions of them, I realized that the interesting part is what came out the other end: something useful, a better decision, a problem solved, or an experiment that taught me to stop.

The token count could keep climbing through all of those outcomes, including the ones that produced nothing worth keeping. It told visitors how much I had consumed without telling them what I had accomplished.

The first replacement was a set of project and build-log counts, which lasted only a few minutes before I decided those could go too.

The projects were already on the page. Readers can open the work, see what it does, and follow the decisions behind it. That is the output I want them to judge. The homepage now gets them there sooner.

Then, while using the site, I noticed it switching from light to dark after I had been reading for a while.

That turned out to be an intentional behavior with an unpleasant result. The site checked the local clock every minute and selected dark mode after 7 p.m. On refresh I was seeing light, followed by a delayed switch to dark. Whatever the original design intent, the experience was jarring.

I removed the timed switching. The site now defaults to dark, remembers a manual theme choice, and leaves it alone while someone is reading.

None of this required a redesign. It required reading the site as it exists today, following the implications of a few changed facts, and noticing what the experience felt like in use.

The build log is becoming useful for that reason too. It gives each project a history, including a place to say that something is finished or that an earlier decision no longer fits.

Tonight, the site became a more accurate account of the work.