About goals, tasks, projects, days, time, and what an AI should and shouldn't do. We built the chain because we believe the chain is the right unit of work. Here's why — explained without hedge.
The chain · one product · one product
Yearly resolutions decay because the horizon is too distant for the nervous system to register. Twelve weeks is short enough that every week feels consequential — 8.3% of the cycle. Not pressure. Pressure gradient.
We chose the 12-week cycle for a single reason: things have to end. Open loops accumulate as low-grade psychic weight; closure is what releases it. Four cycles a year means four chances to retire what isn't working, refresh what is, and start the next thing without the residue of an unfinished previous thing.
A goal is not the unit of action — it's the destination. The unit is the chain that runs from the goal down to what you're doing right now. If a goal can't be traced through a Cycle Outcome, a Weekly Move, and a Today Focus, it isn't a goal. It's a wish.
Instead of
Annual resolutions reviewed in December.
Steady believes
12-week cycles with hard closing dates and a real review.
Every task carries a Next Physical Action — the concrete first step you can sit down and do. "Outline section 2" is a task. "Think about marketing" is not. The brain runs on decisions, and every vague task presents a decision tree the path of least resistance is to defer.
Specificity collapses the tree into a binary: done or not done. No negotiation surface. "Increase revenue 20%" is measurable but not actionable — it presents a thousand paths. "Send 15 personalized outreach messages by Friday" is one path. Pick the one path.
Tasks belong to projects. Projects belong to cycle outcomes. Cycle outcomes belong to vision. Every action knows where it sits, or it doesn't get added.
Instead of
An infinite to-do list with no opinion about what matters.
Steady believes
Tasks with a real Next Physical Action, anchored to a Move.
You go to a project to look or to promote. You don't plan in place. The project shelf holds atomic tasks waiting to be pulled into Today — like a kitchen pantry, not a cookbook.
Every project has a purpose (what it's for, in one line) and a done-when (what proves it shipped). Without those, it's a folder of orphaned tasks that nobody will ever finish. The project header is the contract.
AI does the structuring. Paste a goal — Steady returns 5–15 atomic next-physical-actions. Click "What should I pull into Today?" — Steady reads the project, this week's Move, and today's existing momentum, and proposes 1–3 picks with rationale. The structuring work is the easiest thing for an AI to do well, and the worst use of human attention.
Instead of
Projects as Kanban boards you reorganize on Sunday.
Steady believes
Projects as inventory shelves you pull from on weekday mornings.
The Weekly Move is the answer to: if you ship one thing this week and nothing else, what would advance the cycle? Not three things. One. Pinned to Today, every day.
A week with one Move that ships beats a week with ten things attempted. The brain can hold one concrete commitment all the way to Friday. It cannot hold ten. The Move is the single point of compounding — every day's Focus list anchors to it.
Sunday is for committing. The Reflect + Commit ritual takes five minutes: ship the Move (with a short note that becomes Evidence), drop or carry what didn't ship, name next week's Move. That's the whole ceremony. No journaling marathon. No retrospective theater.
Instead of
A weekly review that takes ninety minutes and accomplishes nothing.
Steady believes
Five minutes on Sunday: ship, log, commit. Done.
Focus is capped at three items. Off-focus work goes to "Other today" — uncapped, but visibly off-arc. Loose ends sit in their own queue, with a 75-minute soft cap. The day already has a shape before you open it.
Three is not arbitrary. It's the number that consistently survives Tuesday afternoon. You'll plan ten. You'll ship three. The cap forces the decision up front, while the prefrontal cortex is online — instead of at 4pm when it isn't.
Loose ends are real. Email, errands, admin, fire-prevention — they aren't Focus, but they aren't optional either. The 75-minute cap is a hard ceiling: more than that and you're using maintenance to avoid the Move. The system flags it. You decide.
There is no infinite list. There is no "backlog of today." There is one Focus list of three, one Other-today list, one capped Loose-ends queue. Open the day, see the day.
Instead of
An infinite daily list that ends Friday with eighty open items.
Steady believes
Three Focus, optional Other, capped Loose ends. The shape is the trick.
Time-blocking doesn't fix priorities — it just gives bad priorities a calendar slot. The question Steady answers isn't "what should I do with my 9am block?" It's: "what should I pull from the chain into Today?"
The calendar is a constraint, not a strategy. It tells you when you're booked. It doesn't tell you what advances the cycle. Steady puts the Move pinned to Today and lets you check the calendar for capacity — not for direction.
Time tracking is also off the table. You don't need a dashboard of how you spent your hours. You need a ledger of what you shipped. One is process theater. The other is the actual record.
Instead of
Hours-spent dashboards and color-coded calendars.
Steady believes
A pinned Move, an honest day, and a record of what shipped.
There's a difference between checking a box and shipping a thing. Steady separates them. Completed tasks fade into the logbook. Shipped Moves become Evidence — a short note that names what was built, when, and why it mattered.
The Evidence ledger is the receipt that proves the system works. When the voice in your head says "you're not making progress," you read the ledger. The voice has a counter-narrative grounded in what actually happened.
Quarterly review reads the Evidence aloud. Not as a brag, as a story: this is what the cycle was. The story is what compounds — not the open-task count.
Instead of
Streak counters, badges, and dopamine-tuned completion animations.
Steady believes
An Evidence ledger that reads like a record of the work, in your own words.
We are not a chatbot you brief. The AI in Steady reads your real context — projects, this week's Move, today's momentum, recent evidence — before it suggests anything. The model is opinionated, not generic, because the system has opinions and the AI inherits them.
Two AI affordances live on the project shelf: Brainstorm (paste a goal, get 5–15 atomic tasks back) and Pull into Today (click once, get 1–3 ranked picks with rationale). Each does the work a thoughtful human would do — but in 8 seconds, not 45 minutes. That's the trade.
The agent never commits without your nod. AI proposes. You promote. The chain stays yours.
Instead of
An AI sidebar that drafts prose and asks you to define an objective.
Steady believes
An AI that breaks down a project into atoms and tells you which to pull right now.
Some opinions are easier to state by what we won't ship.
We reject streak counters. Productivity isn't a Snapchat metric and missing a day is not a moral failure.
We reject "You got this!" cheerleader copy. The voice that helped on day one is a poison pill on day forty.
We reject the infinite to-do list. Every list with no cap is a guilt machine.
We reject configurable productivity systems. The fact that Notion can become anything means it usually becomes nothing. Steady ships with the chain baked in.
We reject yearly goals. Twelve months is the wrong unit.
We reject mood ratings, energy sliders, and emotional check-ins as inputs to the system. The chain holds either way. The AI shouldn't read your feelings before it reads your work.
Instead of
Adding features until users find their own use case.
Steady believes
Removing decisions until the chain is the only path.
One chain holds the work, three things ship today, AI does the structuring, and Evidence beats velocity.