Agentome / Proposed applications

One Agentome platform.
Three application programmes.

Explore three use cases in each programme. Structure-preserving Large Physics Models (LPMs) provide the personal model foundation; backend agents manage the workflow and frontend agents guide users and researchers.

Scientific illustrations use synthetic signals and model states. They show proposed analyses, not participant recordings or validated classifications.

01 / Learning & readiness

Superlearning & Focus

Explore model-informed learning and focus. Frontend agents would guide pacing, practice, and breaks using personal goals and task history.

Application programme / proposal

The research question

Do model-informed suggestions improve learning or task performance compared with task history alone?

Synthetic time-frequency power map aligned to Read, Practice, and Recall periods.
Time–frequency learning profile
01 / PROPOSED

Superlearning Mind Atlas

Compare personal neural-state patterns across learning sessions, with task context and independently measured performance.

Explore the study

Proposed agent workflow

Use the Model Builder and Perturbation Analyst to compare twins across sessions, then ask the Consistency Checker to flag unstable patterns and alternative explanations. Backend agents would manage data quality and the relevant accepted LPM version; continual updates would be evaluated before use.

Frontend guidance

The User Support Guide would explain the atlas and its limits, and collect the learner’s goals and session feedback.

Research inputs

Repeated recordings, learning-session context, and each participant’s reference state.

Proposed research output

A session map with recording context, fit-quality flags, and a report of repeatability. Cognitive interpretations remain hypotheses until tested independently.

What needs evaluation

Repeatability across comparable sessions and prediction of independently measured task outcomes beyond conventional EEG features.

Discuss this research
Synthetic model-feature time course aligned to proposed Practice, Rest, Focus, and Recall periods.
Practice, focus & break timing
02 / PROPOSED

Learning Track Designer

Explore guided lessons, focus sessions, and breaks informed by personal model outputs, goals, and task history.

Explore the study

Proposed agent workflow

Have the Track Designer combine the twin’s state estimate with task history to propose a learning sequence, with its rationale and uncertainty available for review. Backend agents would manage data quality and the relevant accepted LPM version; continual updates would be evaluated before use.

Frontend guidance

The Track Designer would guide the learner through the proposed sequence, explain pacing choices, and offer alternatives using the learner’s preferences.

Research inputs

A learner’s estimated state, task history, and available learning activities.

Proposed research output

A candidate lesson sequence with difficulty and break choices, linked to task history, model estimates, and the policy used to generate it.

What needs evaluation

A prospective comparison with a task-history-only policy using a predefined measure such as retention or task accuracy, with an ablation to test whether EEG adds value.

Discuss this research
Six distinct synthetic signal fingerprints shown on matched time scales.
Recording fingerprints / matched windows
03 / SPECIFIED

Cortical State Library

Organize reproducible recording features into a searchable library, with task annotations and cross-session checks.

Explore the study

Proposed agent workflow

Use the Signal Engineer and Model Builder to connect annotated recordings with twin states, then check whether those descriptions remain stable across sessions. Backend agents would manage data quality and the relevant accepted LPM version; continual updates would be evaluated before use.

Frontend guidance

The User Support Guide would help users explore the catalogue while distinguishing recording annotations from validated mental-state labels.

Research inputs

Consented recordings with task and session annotations.

Proposed research output

A feature catalog with annotation definitions, channel and session context, and a report of cross-session stability.

What needs evaluation

Agreement with independent task labels or measures, with held-out sessions or participants and controls for recording quality.

Discuss this research

02 / Recovery & wellbeing

Restoration & Creative Wellbeing

Explore meditation, music, creative art, sleep, and stress regulation with user-led session choices and personal model context.

Application programme / proposal

The research question

Does model-informed guidance add value beyond fixed sessions and user preferences alone?

Synthetic neural-model feature over time aligned to Meditation, Music, and Creative Art session events.
Session events & model-state profile
01 / PROPOSED

Adaptive Session Engine

Explore guided meditation, music, and creative sessions using preferences, feedback, and evaluated personal model outputs.

Explore the study

Proposed agent workflow

Have the Perturbation Analyst compare state trajectories and the Creative & Restoration Guide organize candidate meditation and audio choices for a controlled wellbeing study. Backend agents would manage data quality and the relevant accepted LPM version; continual updates would be evaluated before use.

Frontend guidance

The Creative & Restoration Guide would help users choose a meditation session, explain timing suggestions, and gather their feedback.

Research inputs

Short research recordings, audio-session events, and a participant’s reference state.

Proposed research output

A timestamped decision log containing available audio choices, the timing rule, model estimates, and the proposal selected for evaluation.

What needs evaluation

Comparison with fixed guidance on a predefined independent session measure, including analysis of missing or poor-quality signals.

Discuss this research
Synthetic sleep-stage temporal profile with Awake, REM, Core, and Deep lanes across an illustrative night.
Awake / REM / Core / Deep temporal profile
02 / PROPOSED

Sleep-Onset Guidance

Study whether estimated sleep-onset transitions can improve audio-fade timing compared with a fixed timer.

Explore the study

Proposed agent workflow

Use the Signal Engineer to check recording quality and the twin to estimate transitions, then propose audio-fade timings for evaluation against a fixed timer. Backend agents would manage data quality and the relevant accepted LPM version; continual updates would be evaluated before use.

Frontend guidance

The Creative & Restoration Guide would explain audio options and collect preferences for the proposed wind-down session.

Research inputs

Research recordings around sleep onset, playback timing, and independent sleep annotations.

Proposed research output

Estimated transition times and proposed fade timings aligned to independent sleep annotations. Sleep-state interpretation requires evaluation against independent annotations.

What needs evaluation

Transition-timing error on held-out annotated recordings before evaluating an audio-timing policy against a fixed timer.

Discuss this research
Before, During, and After synthetic recordings compared on the same time and amplitude scales.
Before / during / after profiles
03 / PROPOSED

Stress Regulation Tracking

Compare neural-state features before and after a practice, alongside independent stress-related measures.

Explore the study

Proposed agent workflow

Have the Perturbation Analyst compare twin states before and after a practice. Use the Consistency Checker and Record Keeper to retain uncertainties and supporting observations. Backend agents would manage data quality and the relevant accepted LPM version; continual updates would be evaluated before use.

Frontend guidance

The User Support Guide would help a user understand the comparison and record practice context and feedback without presenting a model feature as a stress diagnosis.

Research inputs

Repeated before-and-after recordings, practice context, and session-quality information.

Proposed research output

A within-person comparison of fitted features, recording-quality controls, and their association with an independent stress-related study measure.

What needs evaluation

Repeatability, confounding effects, duration of any association, and agreement with independent measures before interpreting a change as stress regulation.

Discuss this research

03 / Longitudinal research

Mental Health & Neurological Research

Support investigator-led mental-health and neurological research with longitudinal model records, personal-reference comparisons, and reviewable hypotheses.

Application programme / proposal

The research question

Do personal model changes add information beyond conventional EEG analysis and independently assessed study outcomes?

Three visits with synthetic signals, coupling matrices, and versioned personal models linked across time.
Longitudinal signals & model versions
01 / PROPOSED

Longitudinal State Record

Follow personal model changes across research visits, with versioned records and acquisition context.

Explore the study

Proposed agent workflow

Use the Model Builder to update the twin across visits and the Record Keeper to preserve recording context, model versions, and the basis for each comparison. Backend agents would manage data quality and the relevant accepted LPM version; continual updates would be evaluated before use.

Frontend guidance

The User Support Guide and Intervention Planner would help investigators navigate the record and explain comparisons using the associated model versions.

Research inputs

Longitudinal research recordings, visit context, and study annotations.

Proposed research output

A visit-by-visit parameter and fit-quality record, with acquisition differences, model versions, and comparison assumptions attached.

What needs evaluation

Cross-visit reliability, missing-data sensitivity, and added information beyond conventional analysis on independently assessed study outcomes.

Discuss this research
Matched wireframe model surfaces compare a personal reference with coupling and delay perturbations.
Reference & perturbed model geometry
02 / PROPOSED

Perturbation Atlas

Compare model changes against a personal reference to frame testable research hypotheses.

Explore the study

Proposed agent workflow

Have the Perturbation Analyst locate departures in twin components, then ask the Consistency Checker to examine confounding factors before proposing hypotheses. Backend agents would manage data quality and the relevant accepted LPM version; continual updates would be evaluated before use.

Frontend guidance

The Intervention Planner would help an investigator explore the atlas, compare assumptions, and frame a testable follow-up question.

Research inputs

Appropriately governed research cohorts with repeated recordings and relevant annotations.

Proposed research output

A parameter-sensitivity map showing stable and unstable model responses, with alternative explanations and follow-up study questions.

What needs evaluation

Robustness across fits, recording conditions, and held-out data. A simulated response does not establish a causal biological mechanism.

Discuss this research
A synthetic network state feeds three counterfactual trajectories for model comparison and investigator review.
Counterfactual model trajectories
03 / EARLY RESEARCH

Neuromodulation Planning

Explore candidate perturbations in the digital twin, with assumptions and uncertainty available for investigator review.

Explore the study

Proposed agent workflow

Use the Intervention Planner and Perturbation Analyst to organize candidate perturbations in the twin, with sensitivity and uncertainty checks for investigator review. Backend agents would manage data quality and the relevant accepted LPM version; continual updates would be evaluated before use.

Frontend guidance

The Intervention Planner would explain candidate model comparisons for investigator review; it would not select or deliver stimulation.

Research inputs

A defined state model, a bounded research question, and explicit perturbation assumptions.

Proposed research output

A comparison of candidate model perturbations with assumptions, uncertainty, and questions for investigator review. It does not select or deliver stimulation.

What needs evaluation

Sensitivity to model fit and assumptions, and whether simulated responses warrant further research. They are not validated predictions of treatment response.

Discuss this research

Agentome / Model-to-experience workflow

Personalize. Evaluate.
Guide. Learn from feedback.

The LPM family provides personal state estimates and forecasting research. Backend agents would keep models evaluated and versioned; frontend agents would turn accepted outputs and user goals into guided sessions. Feedback supports later candidate updates. Application value remains a question for prospective studies.

  1. 01 / LPM CORE

    Personalize the model

    Fit or adapt the relevant personal LPM using eligible recordings and evaluate it on untouched data.

  2. 02 / COMPARE

    Locate the session

    Backend agents compare candidate versions and new sessions with the accepted personal reference before promoting an update.

  3. 03 / PROPOSED

    Guide a session

    Frontend agents explain model-informed suggestions for learning, focus, meditation, music, or art using the user’s goals and preferences.

  4. 04 / PROPOSED

    Record choices and feedback

    Collect user choices and feedback, retain the model version and evidence, and use eligible later data in the next evaluated update.

Evidence & evaluation

An internal prototype.
A reproducible test comes next.

The research account reports an internal comparison on public EEG. This site does not publish dataset identifiers, sample counts, numerical metrics, or the full protocol, so it does not establish comparative performance.

Area Current record Evidence needed
Neural-state tracking Core model research; no tracking benchmark published on this site. Recording protocol, state-feature definition, version, repeatability, signal-quality controls, and independent labels for any mental-state interpretation.
Neural-field forecasting Structure-preserving LPM research; an internal state-model prototype comparison is reported qualitatively. Model-specific datasets, versions, forecast horizons, untouched future sessions, conventional baselines, forecast error, and failure cases.
Continual adaptation / cLPMs Personalization lifecycle in development. Chronological evaluation without future-data leakage, frozen-model comparisons, forgetting and drift tests, update acceptance criteria, and rollback records.
Backend automation Planned operational roles; integrated orchestration in development. End-to-end task success, tool errors, exception handling, model lineage, deployment recovery, and the manual effort required.
Frontend guidance Proposed user support and guided application workflows. Explanation accuracy, usability, user control, unsupported suggestions, and value beyond goals, preferences, and task history alone.
Application benefit Proposed studies and synthetic illustrations. Prospective comparisons with independent learning, sleep, wellbeing, or study-specific outcomes.
Quantum computation A hybrid simulation research direction; no hardware benchmark or speedup result presented here. A defined computational task, execution backend, resource budget, and fair classical comparison before claiming an advantage.
Model fit, physiological interpretation, and application benefit are separate questions. A benchmark on one does not validate the others.

Fitted coupling and delay terms are model quantities. This page does not claim recovery of anatomical brain wiring, and constraint checks do not establish biological accuracy.

Publications in preparation

Read the methods.
Examine the assumptions.

Manuscripts are in preparation. Ask about their scope and current draft availability; they are not presented here as published or peer-reviewed results.

In preparation

Foundation model

The state-model formulation, initial benchmark, and comparisons with conventional methods.

Request the draft
In preparation

Session-variation atlas

The proposed mapping of repeated EEG sessions, task context, and variation relevant to learning research.

Request the draft
In preparation

Longitudinal clinical record

The proposed framework for model-based state descriptions across research visits.

Request the draft

Collaborate with Mindverse

Bring a dataset.
Define a comparison.

Describe your recordings, study question, and current analysis baseline. We can discuss a scoped comparison and the evidence it would need.

Start a conversation

BCI research illustration

Conceptual scientific illustrations with synthetic signals, task labels, sleep stages, model states, and trajectories. These are not participant recordings, validated classifications, confidence intervals, or measured application benefits.