Case study · Product & UX Design
InYourTime is not a task manager. It is a behavioral AI product for the moment when someone knows what they need to do, but the first contact with the task still feels unavailable.
THE ESSENTIAL PRODUCT MOMENTS
Before the AI says anything, the product has already communicated its role: this is not a push, the user can speak naturally, and the system adapts the way in to how the moment feels.
01 · THE PRODUCT PROBLEM
Most productivity products become useful after the user has already chosen what to do and is ready to act. InYourTime is designed one moment earlier.
Reduce the visible field instead of planning the whole thing.
Turn ambiguity into one visible first movement.
Lower contact cost without moralizing avoidance.
“I know what I need to do” and “I can begin doing it” are not the same product state.
02 · PRODUCT LOGIC
The system interprets the real task together with current capacity and blockers, then proposes one concrete entry point — not a plan, not a list, and not a predicted time.
Written in the user’s own language.
Only context that can change the intervention.
Specific, concrete and low ambiguity.
Accept this route or choose another kind of route.
The product must preserve the user’s objective while changing either the friction or the strategy.
03 · BEHAVIORAL ANALYSIS
This is the core of the case study: how the product distinguishes different forms of starting friction and chooses a different behavioral response for each one.
The central behavioral assumption is that failure to start is not one single state. Two users can avoid the same task for completely different reasons, and the same user can need a different intervention on a different day.
So InYourTime does not optimize for “motivation.” It tries to identify the type of friction currently blocking contact with the task, then changes the first move accordingly.
The user knows the task exists, but contact with it is costly, unclear, emotionally loaded, or difficult to sustain.
Energy and blockers are not treated as personality traits. They are temporary context used to select the right intervention.
The system can narrow scope, clarify the first movement, change strategy, lower contact cost, or create regulatory space.
Start step is the meaningful transition. Seeing a suggestion is not counted as action.
The product learns from whether the step felt easier, remained hard, or led to more movement — without turning that history into pressure.
When too many parts are mentally active, the intervention reduces the visible field. The goal is not to persuade the user to “try harder”; it is to make one part cognitively holdable.
When the problem is unclear entry, reducing task size alone may not help. The first step must become observable and specific enough that the user knows what action counts as beginning.
If the user keeps moving away from the task, the system reduces commitment while keeping the objective intact. The intervention is designed to make approaching the task cheaper than continuing to avoid it.
The system should create permission for rough contact: draft, inspect, open, say, sketch. The behavioral target is not confidence; it is reducing the perceived risk of the first move.
Low energy changes how much contact the first step requires. The system avoids assuming a “normal-capacity” day and does not frame lower demand as failure.
The first move should require little working-memory coordination and minimal context switching. The product does not ask the user to build a plan before beginning.
If the user rejects a proposed first step, InYourTime does not treat that as resistance to the product. It treats the rejection as a signal that the intervention targeted the wrong type of friction.
This is why Find Other Way asks what should change before generating another answer.
The system should become better at finding a viable entry point without creating a score, streak, punishment, or identity judgment around the user’s behavior.
Completion can happen later. The product’s first responsibility is to reduce the distance between intention and the first meaningful action while preserving agency.
BEHAVIORAL GUARDRAILS
Opening Active Step is not counted as starting. Start is explicit.
The AI proposes an action, not how fast the user should complete it.
Pause preserves the attempt. Done for now ends it.
Another strategy, less friction or regulation are valid product paths.
No exercise library. One action at a time prevents new decision load.
The seed does not die, decay or shame absence.
04 · THE CURRENT EXPERIENCE
These are the current visual references from today’s version. New states are shown in real mobile proportions instead of desktop cards.
The interface accepts a task, situation or thing the user keeps returning to. It does not ask them to categorize the problem before help begins.
Quick suggestions reduce blank-page friction while remaining editable and non-prescriptive.
A lightweight five-point energy scale is paired with lived-experience blockers. The user can select up to two.
The screen avoids diagnostic language and prevents the state check from becoming another task.
The AI meets the user with clarity and calm, offering one starting point without turning the moment into another decision-heavy screen.
Keep it private and imperfect.
05 · ITERATION
Accepting the suggestion competed with changing it.
Find Other Way now handles the type of recalibration.
The system could count a transition the user had not made.
Only Start step begins timing; returning from pause requires Continue.
Duration can shrink while ambiguity stays intact.
Reduce the contact threshold of the same action, not an arbitrary number of minutes.
The product could not infer intention.
Pause preserves the attempt. Done for now asks Save or Discard.
Accepting, starting, pausing, stopping, changing strategy, lowering friction and regulating are separate product states — not variations of “try again.”
06 · WHAT I WOULD VALIDATE NEXT
Can users predict the difference between a different, smaller and regulatory path?
Does explicit Start step feel clear or redundant after accepting the suggestion?
Can users predict what each option will preserve?
Does every step feel dependent on the actual task?
Does one action at a time reduce load?
Does growth feel meaningful without becoming streak pressure?
CONTACT
Ana Letícia Rezende · Product & UX Designer