Driftline

Research portfolios rarely stall because nobody is working. They stall between stages: a promising signal waits to be sorted, an evaluation finishes without a decision, a program loses its owner, or a field operation reports a blocker that nobody has answered. Those handoffs are difficult to see when evidence, ownership, and status live in separate tools.

Driftline is a fictional decision workspace for Aster Bay Institute. It connects signals, evaluations, programs, and operations in one visible lifecycle, then surfaces the few decisions most likely to restore momentum. A searchable portfolio, people view, and four role-specific lenses let research leaders, program stewards, expedition leads, and specialists see the same work from their own responsibility boundary.

The product does not hide urgency behind an opaque score. Every surfaced item shows the operating rule that triggered it, the evidence on record, elapsed time, current owner, what will clear the condition, and one concrete next action. The interface routes attention; people still decide, delegate, and support.

• A shared record grammar carries research work through signal, evaluation, program, and operation stages.
• Deterministic rules identify stalled reviews, missing ownership, quiet programs, and unanswered field blockers.
• Role lenses reorganize one portfolio around four explicit responsibility boundaries.
• Local demo data makes the complete product flow explorable without an account or external service.
• Light, dark, reduced-motion, forced-color, and responsive states are encoded into the interface contract.

The visual language combines the calm of a research archive with the precision of an instrument panel. Deep ocean fields, pale working surfaces, semantic stage colors, compact Geist typography, and restrained motion make a dense operating model feel deliberate rather than administrative.

I shaped the product premise, lifecycle and responsibility model, explainable routing rules, interaction architecture, visual system, responsive behavior, accessibility constraints, fictional research dataset, and working implementation.