Recommendations

Open Recommendations at /timecampus/recommendations to review suggestions based on your recorded time. Each card includes a rationale and can show the evidence and suggested action behind it. A recommendation is a prompt to review your choices, not a diagnosis or a promise that changing your schedule will improve your wellbeing.

Why am I getting this recommendation?

The four deterministic rule kinds compare the current ISO week with your timeline, targets or previous week:

Rule kind	What triggers it
NO_DEEP_WORK_BLOCK	You have tracked time this week, but your longest uninterrupted focus block is under two hours.
FOCUS_DROP	The previous week has tracked time, and this week's focus ratio is more than 10 percentage points below that week's ratio.
DOMAIN_BELOW_TARGET	You have balance targets, and a domain's actual share is more than 10 percentage points below its desired share. The rule selects the largest shortfall, not one card per domain.
LOW_TIME_WELL_SPENT	Your time-well-spent score is available and below 0.6.

Thresholds describe the recorded data. Missing capture, incorrect domains or a partial week can change the result. Check the evidence before treating a card as a description of your life.

Generation stores at most one recommendation per rule kind, per person and workspace, per ISO week. Running it again in the same week can refresh the evidence, but does not produce another card for the same rule or reset an accepted, dismissed or snoozed card to New. It may produce no recommendations when no rule qualifies or preferences filter every candidate.

The core rules are deterministic. Where optional AI recommendations are enabled for the workspace, additional cards identify their AI provenance separately. The four rules above are not an exhaustive list of those optional AI kinds. You do not need to enable AI recommendations to use the core rules.

Accept, dismiss or snooze

• Accept records the card as Accepted. It does not schedule a focus block, change a target or carry out the suggested action for you.
• Dismiss records it as Dismissed, with an optional reason.
• Snooze records it as Snoozed with a date. The current page chooses seven days ahead and has a separate Snoozed tab. Do not expect generating again to reset its state or create a replacement that week.

The New, Snoozed, Accepted and Dismissed tabs separate those states. Use the suggested action as a starting point for a change you choose yourself.

Control what gets suggested

Recommendation preferences affect generation:

• Muted kinds exclude that kind from new generation. Muting is different from dismissing one card; existing cards remain in their state tabs.
• Opted-in domains restrict domain-specific candidates when the list is non-empty. An empty list applies no domain filter. A candidate without a domain, such as LOW_TIME_WELL_SPENT, is not removed by this filter.
• Quiet hours suppress new recommendation notifications during a UTC hour window. They do not prevent manual generation or remove cards. A window can cross midnight; leaving either end unset, or setting both ends equal, disables the window.
• Never optimise for output is enabled by default. It excludes output-increasing candidates: NO_DEEP_WORK_BLOCK, FOCUS_DROP, and DOMAIN_BELOW_TARGET when the selected domain is Work. It does not mean every recommendation is disabled.

Scheduled summaries and digests have their own opt-in. Merely viewing Recommendations does not opt you into autonomous delivery.

Who can see these cards?

Recommendations and their preferences are self-only within the workspace: the life-time operations use the signed-in actor's identity and workspace, rather than accepting a teammate's user ID. Other workspace members cannot use these views to read your cards or change their state. This is not a claim that every underlying work record is private; time entries and other source features retain their own access rules.

Related: life balance and targets, values and goals, and life timeline.