Harry

If you use AI agents that spawn helper agents (audits, research, big refactors), here's the mechanical fact that changed our setup: the parent never sees the work. The helper runs in its own context; when it finishes, one final message comes back. The forty tool calls behind it the searches that found nothing, the test that failed twice before passing, the edge case quietly ruled out of scope stay behind. What you receive is a summary, compressed by the thing being evaluated.

That's not a bug. Isolation is why delegation is affordable at all. But it means every "Done, all tests pass" is testimony, and nothing in the loop charges the helper for optimism. A caveat is a paragraph; "done" is one word. Guess which survives compression.

Ours got caught the memorable way: an audit subagent confidently reported that a sourced claim had no source behind it. The source existed. We found the exact sentence in the original docs and ended the argument in minutes.

What we require now, roughly in order of payoff: evidence formats written into the dispatch prompt (every finding arrives with a verbatim quote and a file path), artifacts over adjectives (test output and re-runnable commands instead of "thoroughly checked"), and spot-checks scaled to blast radius anything we'll act on gets one finding re-derived end to end. The re-check is deliberately cheap: when a claim comes holding a quote, verifying it is a text search.

It's the first question friends asked when I showed them https://convert.softorino.com, and it's fair, free converters usually pay for themselves with ads, tracking, or your files. The honest answer: the catch is that some of you will like it enough to look at the paid Mac apps I work on. That's the whole model, a good free tool as the introduction. Launch is this Sunday (Aug 30), so before then I want to hear it straight: what makes you trust a free tool, and what instantly kills that trust?

I ship one app to seven surfaces on my own: web, iOS, Android, CarPlay, Android Auto, Fire TV and Alexa. I did not plan that. Each one looked like a small addition to something that already worked, and each one turned out to have its own review queue, its own release cadence and its own way of breaking quietly.

Three things I did not price in. The first is that a platform you add is a platform you keep testing forever, and the failures are invisible from your desk. Both my mobile apps were locked to portrait, which meant they were phone apps pretending to be tablet apps for months. I only found out when I finally ran them on a tablet. Nothing was broken, nothing was reported, they were just quietly bad on a whole class of device.