I’m curious.
And I like
trying things.

I’m experimenting with agents that plan and run security tests in authorized environments. I’m as interested in what they can do as in how to check their conclusions and spot when they’re wrong.

In these experiments, finishing a task and proving a finding are two things I check separately.

That question led me to try models on a phone and combine their answers with saved documents. There’s an Android app in testing, but the exploration also means comparing models, response times, and what’s actually useful without internet access.

I’m still evaluating answer quality. What works on one phone won’t necessarily work the same way on another.

I’m experimenting with a personal monitor that brings sources together, groups topics, and uses AI to summarize and flag relevant items. I’m still exploring how to preserve context and catch summaries that leave out something important.

There’s a working monitor for now. Other ways to investigate and test hypotheses are still in the notebook.

I’m trying ways to observe BLE, Wi-Fi, and cellular signals from a phone. I compare what appears in the captures with what I can infer from them, using my own tools and open-source components.

A signal can offer a clue without being enough to identify its source. The cellular experiments are still limited to one specific device.

I’m trying a path from transcripts to articles and references I can return to later. Agents help organize and write; then I check what they kept, what they left out, and whether they stayed faithful to the source.

The method and the source material are separate things. Here I share the process, using my own examples or material I have permission to use.