Priority depends on more than urgency.
Intent, feasibility, time, place, people, and momentum all influence what should surface.
Selected work / Product 05
Context-aware prioritization
Stratum
A mobile prioritization engine that continuously re-ranks work against the user's present reality—available time, location, people, resources, intent, and momentum.
Mobile product capabilityA priority feed that can explain why the order changed.
The yellow-and-black mobile interface makes the ranking model visible: work rises or falls as time, location, people, available resources, urgency, and recent momentum change around the user.
The product problem
A static list knows what exists. It does not know what is executable now.
Most task managers preserve user-entered order or a simple due-date sort. They cannot account for a fifteen-minute gap, the person currently nearby, the resources available, the user's scheduled location, or the cost of switching away from work already in motion.
Stratum treats priority as a dynamic score rather than a permanent label. Every task can rise or fall as context changes, creating a feed that answers a more useful question: what is the highest-leverage action I can realistically complete now?
Work exists inside a graph.
Tasks connect to projects, areas, subareas, people, locations, schedules, and required resources.
Dynamic ranking must still feel immediate.
Heavy graph computation cannot turn every feed refresh into an expensive multi-table query.
Product model
From backlog to a contextual priority router.
The product combines deliberate user intent with environmental feasibility. Work is not arbitrarily hidden; it moves between prominence bands as its context changes.
Hierarchical work model
Tasks and subtasks connect to projects, areas, and subareas without flattening the user's broader commitments.
Context vectors
Time required, due dates, manual importance, resource constraints, location, people, and momentum contribute to dynamic prominence.
Spatial and relational graph
Nested locations and many-to-many relationships model where work can happen and who meaningfully affects it.
Schedule-aware feasibility
Schedule blocks establish available windows, domains, locations, and nearby people for the ranking engine.
Dynamic prominence feed
Mission-critical work, relational boosts, contextual triggers, and deferred tasks remain visible in understandable bands.
Fast capture surfaces
Focused sheets support rapid creation and editing of tasks, domains, people, locations, and the metadata that changes rank.
Architecture
A graph-relational model with a dedicated speed layer.
The normalized relationship graph protects flexibility, while precomputed context gives the mobile feed a fast, stable read path.
Context computation
Entities and polymorphic edges preserve rich relationships. PostgreSQL triggers flatten ranking inputs into a JSONB compute cache so the feed does not reconstruct the full graph on every refresh.
Polymorphic graphPostgreSQL ltreeJSONB cacheDatabase triggers
Mobile foundation
A Flutter client, Riverpod state layer, Supabase backend, role-aware data access, scheduling integration, and local persistence foundations support a responsive cross-platform product.
FlutterDartRiverpodSupabasePostgreSQLDrift direction
Prioritization that responds to life as it is happening.
Stratum's value is not another place to store tasks. It is a decision layer that continuously reconciles ambition with the constraints and opportunities of the current moment.
Build an intelligent workflowNext product
Strata Timetable →