A single-file, offline-first workspace that unifies tasks, follow-ups, and links across a multi-product portfolio, built because no enterprise tool treats "who am I waiting on" as a first-class thing.
Running a multi-product portfolio means a different work-item board, doc space, and stakeholder cast for every product. None of it aggregates at the personal level. Follow-ups mostly lived in email flags and memory instead, and some went stale. I also had about 70 bookmarks I'd stopped using.
Two weeks of structured self-observation surfaced the clearest finding: follow-up work has a distinct mental model from task work. It needs an owner and a source, and its own review cadence. No tool at the personal level modeled that.
"Capture friction is the whole ballgame. If the tool of record demands too many fields, the thought just doesn't get captured."
I made follow-ups a first-class type, with an owner, a source, and its own board, rather than a tag bolted onto a task. And I built the whole thing as a single-file, offline, browser-storage app with zero backend, specifically so it needed no security review, no hosting, and no approval to start using immediately.
The trade-off, stated honestly: no cross-device sync, a hard storage ceiling, and no framework, in exchange for a tool that got used starting day one instead of never shipping at all.
The cheapest tool that ships beats the ideal tool that doesn't. Zero-approval, zero-infra architecture was the actual reason this exists in daily use.
Model the user's mental model, not the domain's vocabulary. "Blocker" and "dependency" felt like statuses but were really reasons.
Watch your own behavior for a week before designing the data model. A proper browser database from day one would have avoided the storage ceiling.
Parsing supplier PDFs with LLMs and business rules.