About Focuson
Focuson is built around a few decisions that are easier to explain as a story than as a feature list. Here they are.
Why the day has a ceiling
When a task list is long, what happens is not "a lot of work" but paralysis — volume itself triggers a stop. The day plan has a capacity: roughly 3, 5 or 8 tasks depending on how full a day you named. An hour of buffer for the unexpected is already subtracted.
It is not a hard wall: ten five-minute tasks fit where three hour-long ones will not. And we never show "42% done" — the number that matters is what got done, not the distance to a bar.
Why the assistant is quiet
The assistant speaks at most once a day and never counts your misses. If it has something to say, it waits until you open the right place yourself. No daily nudges, no model initiative.
Every line it writes is checked by a separate layer for reproach, lists and needless length — and rewritten if something is off.
A portrait, not a dossier
We accumulate understanding — you, your spheres, your projects — as digested, typed facts, not a chat log. "Underestimates time." "More productive in the morning." You can open and correct them.
In 0.7 the portrait also changes what the system does: planning facts like "do not plan in the evening" are read deterministically when the day is assembled. That is adaptivity — as a readable fact, not a hidden weight adjustment.
Why there is no methodology switch
One system for everyone. The difference between people lives not in a chosen "mode" but in what the system understood about you. Choice is a tax, and we do not levy it where we can avoid it.
If you know GTD — capture, organise and a weekly review are here; there are no mandatory per-task contexts. Someday/Maybe exists as a distinct task state.
Carry-over with dignity
Whatever you didn't finish yesterday is marked as carried over: it does not dissolve into the list and it does not vanish. You see the debt — and you understand it, because a reason code sits next to it.
Revising a decision is normal, not undoing a mistake: day rebuilds can be undone with one button.
AI is a layer, not the core
The model never rearranges anything silently and never owns the rules. Dates, priorities, the ceiling — that is the engine. AI phrases, suggests and explains "why this task now", sitting inside the interface rather than replacing it.
The daily loop works with AI off: every feature has a deterministic path.