This is the first piece here — a statement of intent, before the actual articles arrive.
Two meanings
"lostcontext" means two things, and both are on purpose.
Technically, the name describes a real phenomenon. Every large language model works with a context window: the slice of a conversation it can process at once. Fill that window and older parts drop out — the model loses the thread. The gradual decay in quality that follows even has a name in the field: context rot.
Socially, the name means something else. The public debate about AI keeps losing its context, too. It tips between salvation and doom and skips the concrete middle: what these tools can really do, where they help, where they fail, and why. That's the context I want to hand back.
From the inside, not the sidelines
I don't write here as a spectator. I build applications — and I use the very AI tools I write about while doing it: daily, in real projects, with real mistakes. That vantage point is the heart of lostcontext.ai. Not "an expert says", but "here's how it went when I built it".
What grows here
Two things. First, articles that show how these systems work and where their limits are — calm, concrete, no hype. Second, a Lab of small interactive pieces that make an abstract idea something you can touch.
What won't happen here: no scaremongering, no superlatives, no numbers without a source. Where I can't back something up, I say so — or leave it out.
Transparency
Right now this site earns nothing and costs no one anything. Should that change — through restrained advertising, say, or the odd recommendation link — I'll announce it here, openly, before it happens. No hidden tracking, no bought opinions.
An invitation
If you've read this far: glad you're here. This place grows slowly and on purpose. Come back — or write to me if something's on your mind.
/compact — the essentials, if context is running low:
"lostcontext" means both: language models lose context (context rot) — and so does the public AI debate. This site is written by a practitioner from the inside, calmly and with no numbers without a source. What's growing here: articles, and a lab that makes abstract ideas tangible.