TRACE
TRACE is the bridge between perception and memory: Roughly 60% perception and 40% memory. It is not the memory system itself. That role belongs to OMM.
CYGNUS extracts signals. ORACLE detects patterns. LUCID interprets them. Together, they produce a detailed behavioral picture of a single interaction. But what happens when that person comes back tomorrow? Next week? Next month?
Without cross-session context, each session starts from zero. TRACE carries distilled findings forward and builds the personal baseline that connects perception across time.
TRACE doesn't re-analyze old data. He doesn't store raw signals or video frames or audio recordings. What TRACE stores are the distilled findings: ORACLE's pattern detections, LUCID's interpretations, behavioral metrics, and session-level summaries. Over time, these accumulate into a behavioral profile that grows richer with every interaction.
TRACE connects sessions into a continuous behavioral narrative
A guided audio version of this page, adapted for clarity and flow rather than read word for word.
TRACE connects sessions into a continuous behavioral narrative
Most behavioral analysis systems operate in a single time window. They analyze one video, one call, one session, and produce results. The next session is independent. There's no thread connecting them.
TRACE adds that thread. He connects sessions into a continuous behavioral narrative. When you review a TRACE-enhanced analysis, you aren't just seeing what happened in this session. You're seeing how it compares to previous sessions. What's changed. What's stayed the same. What's trending in a direction.
This distinction matters enormously in practice. A student who shows hesitation signals in a single tutoring session might be having a bad day. A student who shows increasing hesitation signals over six weeks is experiencing something systematic. A teacher reviewing a single session can't tell the difference. A teacher reviewing TRACE data can.
Collection first. Synthesis second.
Session-Level Metrics
Aggregated behavioral data from CYGNUS. Average Action Unit activations, pitch trajectory, speech rate patterns, postural dynamics. These aren't raw per-frame values but session-level statistical summaries: means, ranges, distributions, trends.
ORACLE Findings
The complete set of pattern detections from the session. Which rules fired, at what timestamps, with what confidence, and how they clustered.
LUCID Interpretations
The contextual readings of the session's behavioral data. Key observations, cross-modal insights, and identified limitations.
Session Metadata
Duration, configuration used (Standard, Lite, ECHO), channels active, and session identifier.
Baseline Computation
TRACE calculates rolling behavioral baselines from all previous sessions. What's this person's typical pitch range? Their average speech rate? Their characteristic postural dynamics? These baselines evolve over time as more sessions accumulate, and they're always computed from the individual's own data. Population averages play no role.
Trend Detection
TRACE identifies longitudinal trends in the behavioral data. Is the person's vocal variability increasing over the past five sessions? Is their postural openness decreasing? Are hesitation clusters becoming more or less frequent? These trends are computed statistically and flagged when they reach significance.
Deviation Alerts
When the current session's data deviates significantly from the established baseline, TRACE generates an alert. "This session's average pitch is 18% higher than the personal baseline." "Hesitation cluster frequency is double the average of the last 10 sessions." These deviations don't carry interpretive labels. They're quantitative observations that give ORACLE and LUCID additional context for the current session.
Progression Tracking
For ongoing relationships, TRACE tracks behavioral progression over time. How engagement metrics have evolved. How confidence indicators have shifted. How cross-modal coherence has changed. This progression data is available at the individual level and, when aggregated with consent, at the group level.
What only becomes visible over time
Recurring Divergence
If cross-modal divergence appears consistently when certain topics are discussed across multiple sessions, TRACE flags this as a recurring pattern. The pattern exists at a level that no single-session analysis can detect.
Behavioral Seasonality
Some behavioral patterns follow rhythmic cycles. Weekly patterns, temporal patterns, and contextual patterns. TRACE detects these cycles when enough data has accumulated.
Gradual Shift
Changes that happen so slowly they're invisible within any single session but become clear over weeks or months. A person who's gradually becoming less vocally expressive, or whose hesitation patterns are shifting in character. TRACE's longitudinal view makes these gradual shifts visible.
Regression Alerts
When positive behavioral trends reverse. If someone who's been showing increasing engagement metrics over months suddenly starts declining, TRACE flags the reversal. Early detection of regression is one of his most valuable capabilities across educational, clinical, and professional contexts.
He compares each person to themselves, never to others
TRACE's baseline is strictly personal. He compares each individual to themselves, never to others.
This is a deliberate design decision with significant implications. Population-average baselines are convenient but fundamentally flawed for individual analysis. A naturally quiet person who speaks at 100 words per minute isn't disengaged just because the population average is 140. A naturally expressive person who shows high gestural activity isn't agitated just because most people in similar settings are stiller.
TRACE builds his baseline from the individual's own behavioral data across sessions. The first few sessions establish an initial baseline. Each subsequent session refines it. Over time, the baseline becomes a nuanced, multi-dimensional portrait of how this specific person typically presents across the behavioral channels.
Deviations from this personal baseline are far more meaningful than deviations from a population average. When a person who typically speaks at 100 words per minute suddenly drops to 70, that's a significant personal deviation. When a person whose facial dynamics are typically varied becomes unusually still, that's notable for them. TRACE detects these personal-scale changes precisely because he knows what's normal for this individual.
CANON, the trust and calibration framework, operates in a similar philosophical space: personal over population. But the two are independent. TRACE tracks longitudinal patterns across all behavioral channels over weeks and months. CANON calibrates the vocal baseline progressively through its five layers. They don't depend on each other. TRACE works without CANON, and CANON works without TRACE. When both are active, they complement each other: CANON provides a deeply calibrated vocal reference point, and TRACE provides the longitudinal trajectory across all channels. But neither requires the other to function.
Built from the individual's own sessions. Refined over time. Never population-ranked.
A drop from 100 words per minute to 70 matters because it is unusual for that person.
Philosophically aligned, technically independent. Together they deepen calibration and trajectory.
TRACE stores the analysis, not the source
Session-level metrics, ORACLE findings, LUCID interpretations, computed baselines, detected trends, session metadata. All stored in structured format, linked to the individual's profile.
Raw video frames, raw audio, frame-level CYGNUS data, or any original media from the sessions. TRACE receives the analysis, not the source.
TRACE data is Attributed Data as defined in the Privacy Policy and Data Processing Agreement. It belongs to the individual and/or the deploying institution, depending on the deployment agreement. EXIDEUS LLC processes this data on behalf of the data controller under the terms of the DPA.
Retention is configurable by the deploying institution. Individuals can request deletion under GDPR rights. Access is limited to authorized roles defined in the deployment agreement.
Longitudinal value appears wherever people return
Education
A teacher reviewing a single session with a student sees a moment in time. A teacher reviewing TRACE-enhanced data sees a trajectory. They can see whether the student's engagement is building or fading, whether confidence in specific topics is growing, whether behavioral patterns suggest the student is struggling with something that hasn't surfaced in grades or test results yet.
Behavioral regression often precedes academic regression. A student who starts showing increased hesitation and decreased vocal variability over several sessions may be headed toward disengagement well before it shows up in their academic performance. TRACE makes this visible.
TRACE also enables meaningful progress tracking that goes beyond grades. A student whose behavioral engagement has measurably increased over a semester, whose hesitation patterns have shifted toward shorter, less frequent pauses, is showing real progress even if their test scores haven't moved yet. TRACE provides evidence of behavioral development that traditional metrics can't capture.
Sales and Client Relationships
Sales teams working with the same prospects over multiple meetings build TRACE profiles that reveal how the relationship is developing. Is the prospect becoming more open over time? Are the hesitation patterns around pricing decreasing or increasing? Is the vocal engagement growing with each call? TRACE turns a series of disconnected sales calls into a readable relationship trajectory.
Coaching and Speaker Development
A coach working with a speaker across months of practice sessions can track exactly how their delivery is evolving. Pitch variability increasing over time. Pauses becoming more deliberate rather than hesitant. Cross-modal coherence improving. TRACE provides objective evidence of development that the speaker themselves might not notice, and it flags regression early enough to course-correct.
Clinical and Therapeutic Settings
In therapy, TRACE tracks behavioral patterns across sessions in ways that complement clinical observation. Gradual shifts in vocal quality, evolving pause architecture, changing engagement patterns become visible against a longitudinal profile. A therapist reviewing TRACE data can see trends that are invisible within any single session.
Research
Longitudinal behavioral research requires exactly what TRACE provides: consistent measurement of the same individual across time, with personal baselines and statistically validated trend detection. Researchers can separate individual trajectory from noise, track treatment effects over time, and detect regression or plateau with precision.
TRACE only becomes valuable when sessions can accumulate
TRACE requires multiple sessions with the same individual to produce meaningful longitudinal data. His value scales with the number of sessions: more interactions create richer baselines, more reliable trend detection, and more nuanced deviation alerts.
For this reason, TRACE is most valuable in ongoing relationships: educational programs, coaching engagements, clinical sessions, research studies. In one-off interactions, TRACE provides limited value beyond basic session archival.
TRACE also requires that sessions be linked to the same individual. This linking happens at the application level, not at the TRACE level itself. TRACE receives data tagged with an identifier and builds the profile accordingly. He doesn't perform identity verification.
Institutions that deploy TRACE must have appropriate data processing agreements in place, as TRACE data is Attributed Data under GDPR. The standard Data Processing Agreement covers TRACE deployments.
The practical questions that decide whether memory is usable
TRACE starts providing longitudinal context after as few as three sessions. The initial baseline is rough but functional. After 10+ sessions, trend detection becomes statistically reliable. After a semester's worth of regular sessions, TRACE's behavioral profiles and trend analyses reach their full depth.
Operational boundaries in one longitudinal frame
TRACE is the longitudinal tracking layer of the OPM pipeline. For information about how TRACE receives his input, see ORACLE and LUCID. For information about personal vocal baseline calibration, see CANON. For how perception data is used across products, see the Privacy Policy.