Feature
Signal Insights
Signal Insights v2 is Diorama's component-based insights experience for exploring behavior and health signals. It adapts detail sections by signal type so each page shows the most relevant analyses.
Tasha used to open Insights and feel swamped — every page looked the same but meant something different. With detail tailored to each signal, her sleep, mood, and activity pages each lead with what matters for that one thing. She spends less time reading charts and more time changing something.
Why it matters
One-size-fits-all charts can hide meaning when different metrics behave differently. Tailored signal components improve interpretation speed and help users find useful patterns faster.
How Diorama uses this
Browse categorized signals from the Insights tab. Open detail views that dynamically show components like trends, averages, variability, distribution, and baseline deviation. Use streak and adherence-oriented sections for habit-focused signals.
What you can do
Insights v2 lets you inspect each signal with component blocks chosen for that signal's archetype and supported analyses. This means high-volume metrics can emphasize trend and variability, while habit-like metrics can emphasize consistency and completion behavior.
Signal detail pages combine visual sections and computed stats so you can move from "what happened" to "what pattern might matter" without leaving the app.
The same architecture also supports incremental expansion, so new insight components can be added without rebuilding the whole experience.
Notes and limitations
Component visibility depends on data quality and sample size, so not every signal page will look equally dense.
A few advanced metadata pathways are already modeled but not fully surfaced in production UX yet.
The science
How information is shown should follow what it means. A view built for the specific signal strips out the irrelevant and foregrounds the change worth noticing, and the less work it takes to read an insight, the more likely it is to be acted on rather than abandoned.
Limitations
Some components are hidden on sparse datasets, so depth can vary by signal. Event-focused signals still have fewer specialized surfaces than some continuous metrics. A few metadata-driven capabilities, such as deeper ai-context narration and event-frequency specialization, are still incomplete.
Understanding yourself is often a design problem before it is a discipline problem. When the pattern is easy to read, change stops feeling abstract — and that is usually what makes consistency reachable.
Diorama
See your whole picture.
Diorama brings training, recovery, and health together so you keep competing — not just keep going.