Autonomic Intelligence

How our data, neuroscience and machine learning work together

What Autonomic actually measures

Autonomic follows how an individual's focus, mood, energy, sleep and stress change over time. Users complete a roughly three-minute daily check-in, and the platform builds a longitudinal picture of that individual rather than comparing a single observation against a generic population.

  • Focus
  • Mood
  • Energy
  • Sleep
  • Stress

What our models are trained on

Our machine-learning models were developed using Autonomic's own longitudinal dataset, collected over five years from people using the platform. The dataset now contains more than 350,000 records. The models learn relationships between repeated measures, behaviours, interventions and outcomes over time.

Training population
910 participants
Human-labelled training data
~205,000 responses
Connecting longitudinal signals to assessment and intervention selection
Average observations/person
~320 data points
Time represented
5.5 years
Performance
85% accuracy

Targets

  1. 01Next information to gather
  2. 02Relevant pattern or change to surface
  3. 03Appropriate intervention to recommend
  4. 04Need for further assessment
  5. 05Need for human review

Inputs

  • Longitudinal focus, stress, sleep, mood & energy
  • 18 proprietary survey instruments
  • Binary (Yes/No) and 1–10 scaled responses
  • Free-form text
  • Behaviour and intervention history
  • Engagement and response patterns
  • HRV, voice & facial signals, EEG*

*Objective and multimodal signals are progressively being integrated into Autonomic's intelligence layer.

The Autonomic data loop

Every interaction can capture three connected layers

  1. 01 · Signal

    What is happening

    Behavioral and physiological indicators associated with brain performance

  2. 02 · Decision

    What Autonomic does

    What was assessed, recommended or assigned, including validated human decisions

  3. 03 · Outcome

    What happened next

    How the individual's signals and behaviour changed following the decision

Separating the model from the intervention

The model identifies patterns in an individual's longitudinal data. Autonomic's decision intelligence determines what is relevant and which approved response is appropriate, based on confidence thresholds, safety rules and the individual's history.

Recommendations are drawn from Autonomic's proprietary neuroscience knowledge system, which translates established research into approved insights, behavioural strategies and evidence-based actions. Human oversight governs adaptation and provides an additional layer of validation and safety.

Autonomic has several different things happening:

01

Measurement

What an individual reports subjectively or is captured objectively.

02

Machine learning

What patterns Autonomic identifies.

03

Decision intelligence

How Autonomic determines what information/action may be appropriate.

04

Intervention

The neuroscience-based habit/action/message delivered.

05

Human validation

Where trained validators oversee/adjudicate outputs.

What does our research show?

Study 1

Behavioural

N
376 Autonomic users + 50-person non-intervention comparison group
Population
Undergraduate students from 5 universities and colleges
Duration
10 weeks
Measures
Focus, mood, energy, sleep and stress

+29.1%

Sleep

+24.9%

Mood

−17.6%

Stress

+13.1%

Focus

+10.5%

Energy

All five changes within the Autonomic group were statistically significant (p < .001). The comparison group showed no significant pre-to-post changes.

Because this was a nonrandomized study, the findings demonstrate significant improvements associated with Autonomic use, but do not by themselves establish causation.

Study 2

Behavioural + Biomarker + EEG

N
30
Population
University of Victoria undergraduate students
Duration
10 weeks
Measures
Behavioural measures + cortisol + dopamine + EEG during cognitive testing

+24.78%

Focus

+19.32%

Energy

+15.39%

Sleep

−23.6%

Cortisol

−7.9%

Frontal theta during working memory

−13.6 ms (−3.3%)

P300 latency

Focus and energy improved significantly over time, and sleep showed a significant linear trend.

The EEG findings showed significantly reduced frontal theta during working memory and significantly faster P300 latency, consistent with reduced cognitive effort and faster information processing during the task.

Cortisol decreased by 23.6%, but the change was not statistically significant. We report it as an observed biomarker change, not as evidence of a confirmed cortisol effect.

What we don't know yet

Our evidence is strong within the populations we have studied. We do not assume those findings automatically generalize to everyone. Autonomic has been scientifically studied in university populations, with published research demonstrating measurable changes across behavioural, physiological and neurophysiological measures.

We have not yet established Autonomic's efficacy in:

  • People under 18
  • Adults over 65
  • Neurodivergent populations
  • Other populations not yet adequately represented in our research

We are continuing to expand the populations we study, investigate which components of Autonomic contribute to observed outcomes, and evaluate how objective measures such as EEG, HRV, facial movement, voice, and passive signals can strengthen personalization over time.

Where the evidence is established, we will say so.Where it is still emerging, we will say that too.