Shown with sample data
Turning an overloaded day into a realistic, weighted plan.
A to-do list that weighs your day. I designed the experience and directed AI agents through the build.
- RoleDesign direction, product decisions
- TimelineSept 16-24, 2026
- StatusLive web app, redesigned October 2026
I set the product rules and interface direction, reviewed builds, and chose what shipped. AI agents handled implementation and the documented QA pass.
The problem
Ten errands and one difficult project both count as tasks, though they take different amounts out of a day. A list looks manageable until the work that matters has nowhere to go.
I redesigned around one constraint: weight. Light tasks are 1 point, Medium 2, and Heavy 3. Extra tasks count as zero because they’re optional, not commitments. A load meter warns before the day gets too full.
Weight is not time
A short but demanding task can still be Heavy. Separating effort from minutes keeps the load meter from becoming another time tracker.
Optional stays optional
Extra carries zero points, so an optional task doesn’t compete with the day’s commitments.
How it came together
Between September 16th and 24th, I reviewed builds against a simple rule: the day should show what matters without making every task look equally demanding.
Weight limit on the day
The load meter shows when the day is full. Light, Medium, Heavy, or Extra. Primary and Secondary priorities sit on top so the most important work is the first thing you see.
AI is optional
Gemini can order the day into time blocks if you allow it. When it is off or out of quota, an on-device fallback keeps planning available.
Nothing moves silently
Plans and end-of-day check-ins only suggest. Nothing changes until you apply it.
Four views
The October redesign split Stellar into Dashboard, Tasks, Habits, and Settings. Desktop gets a sidebar you can collapse, and phones get a bottom tab bar.
Shipped state
Version 1.8 ships without requiring an account. Google sign-in adds sync. Gemini planning starts only if you opt in.
Before release, I directed an agent to use the app like a new user. It found ten issues, including habit check-ins on future days and tap targets under 44 points. I directed fixes and an agent-run verification pass at desktop and phone widths.
The next step is to compare planned time with actual time, then test whether that feedback makes the workload estimate more useful. Native App Store packaging remains future work.
An agent-run QA pass identified ten issues before release. These screens show the shipped interface with sample data, not evidence of recruited-participant usability testing.
