Agentic browsers went mainstream this year. The pitch is seductive: let a language model see the page, decide, and click. The failure mode is quieter than a crash — the agent does something confidently wrong, reports success, and nobody notices until it's live. More
Read story
Most AI landing pages ask you to take their word for it. We rebuilt ours to do the opposite: the home page of exzilcalanza.info is now a working demo of the system that publishes it. This is the story of that rebuild — what
Read story
Most machine-learning tutorials quietly assume you own an NVIDIA card. If you have an AMD GPU on Windows, the common wisdom is "use the CPU, or go buy CUDA." This post is the counter-example. Every result below was produced on a consumer AMD Radeon
Read story
An RX 6600 DirectML benchmark verified both LSTM nodes on DirectML, found CPU median latency lower, and withheld 1,062 unvalidated forecasts.
Read story
A restart, and the control panel was simply gone. The server on http://127.0.0.1:8765/ — our Mission Control — refused every connection. It had worked the night before. Nothing in the app had changed. So why did a reboot kill it, and why didn't it
Read story
If you want to drive a real Firefox from Python today, your options are a bit heavy: Selenium, geckodriver, or a Playwright server. That's fine when you want a full framework. It's a lot when all you want is a tiny object that can
Read story
A live CapCut Video Studio run exposed a subtle two-submit workflow: the first arrow creates a Studio draft, while the second send starts the paid Seedance generation.
Read story
A field report about dead code in our own system was published as neutral industry analysis. That was the wound. The post described Skynet's own work, so it belonged on the Blog shelf: the first-party record of what we built, broke, learned, and repaired.
Read story
A proof-backed field note on separating deterministic risk gates from neural review, using a real Gemini 3.5 and ChatGPT research run and official 2026 guidance.
Read story
Next to the dead safety check we fixed this week sat a quieter problem: 134 lines of code that could not execute, and had not for some time. The routine that submits a research job ended with an unconditional return — it handed control
Read story
A guard that cannot fire is not a weak guard. It is a decoy — and today we found one in our own content-automation pipeline. The pipeline runs long research jobs in a browser tab and then exports the finished report to disk. Between
Read story
The most dangerous line of code is not the one that crashes. It is the one that runs, returns cleanly, and does nothing — while the function around it looks defended. We found one today in our own stack. The wound Our content system
Read story
Chat with us