Enactive AI · Collaborative temporal science

Kalyriel Scope

A Collaborative Temporal Science Instrument

An Open Platform for Discovering, Interpreting, and Sharing Temporal Knowledge. Kalyriel Scope combines continual computational discovery with human semantic interpretation to enable collaborative exploration of dynamic systems. By discovering temporal motifs (recurring patterns that unfold through time), it supports expert annotation, and enables shareable motif libraries. The platform helps scientific communities build evolving knowledge about temporal phenomena across domains such as healthcare, neuroscience, climate science, industrial monitoring, finance, and human–AI interaction.

Why this is different

From Prediction to Temporal Explanation

The Emergence Machine does not only predict a signal. It organizes the signal into an interpretable temporal structure—identifying phases, transitions, recurring motifs, disruptions, and recoveries—and renders that structure as an evidence-grounded report.

Kalyriel ScopeAn open platform for discovering, interpreting, and sharing temporal knowledge.
No motif library loaded
Local report
Ready with a generated demonstration signal.
Current regime
—
Waiting for samples
Attractors L / R / G
0 / 0 / 0
Established recurring structures at three temporal scales
Regime F1
—
Detected regime events vs. high-drift anomaly states
MAE
—
Mean absolute one-step forecast error
MSE
—
Mean squared one-step forecast error
Cross-scale coherence
—
Agreement among local, regional, and global models
Tiny EM adaptive forecaster
A causal one-step predictor adapts its learned local velocity after every realized error.
Level 0.00 Trend 0.00 Variance 0.20 Momentum 0.75 EMA error —
Adaptive plasticity 0%
Surprise and changing residual direction increase adaptation and reopen attractor learning.
Current forecast skill vs. persistence
—
Rolling-window skill. Positive values outperform persistence.
Session median skill—

Signal, adaptive prediction, and regime boundaries

The one-step-ahead prediction is plotted at its target sample and learned online from realized error.

signalpredictionshiftannotation

Regime formation through time

Each color represents a distinct system configuration inferred online.

Drift pressure

Rising pressure means the current pattern no longer fits the active attractor.

System interpretation

A plain-language summary of what the model is doing now.

Load a demo or upload a numeric data column to begin.

Local attractor landscape projection

Fast patterns · projected by mean and slope · size = occupancy · opacity = recency

Regional attractor landscape projection

Intermediate patterns · projected by mean and slope · size = occupancy · opacity = recency

Global attractor landscape projection

Slow patterns · projected by mean and slope · size = occupancy · opacity = recency

How to read the Emergence Machine

1 · Observe
The machine receives one value at a time from a CSV column or generated signal.
2 · Form attractors
Recurring local patterns become stable centers rather than being treated as isolated points.
3 · Detect regimes
Persistent changes in attractor occupancy and drift become regime transitions.
4 · Adapt
Forecasts are rebuilt from the currently active structures across three timescales.
Kalyriel Scope: No data leaves this browser. This open research platform makes temporal motifs, semantic annotation, motif libraries, and adaptive dynamics visible and explorable. It is a research prototype, not a validated production forecasting or clinical decision system.