Representative human brain formed from silver-blue neural fibers and nodes, with EEG-like recording traces

Company & investment

Continual physics models.
One Agentome platform.

Mindverse is developing Agentome around structure-preserving Large Physics Models (LPMs). Continual personalization would keep individual models aligned with incoming BCI data; backend agents would operate the technical lifecycle, while frontend agents guide users in learning, focus, creative wellbeing, and research. These capabilities need separate evidence and partner evaluation.

The investment rationale

One continual model core.
Two agent layers.

The platform hypothesis is that one personalized model foundation can support multiple guided experiences while backend agents reduce manual technical work. Model accuracy, continual adaptation, operational reliability, and user benefit must each be evaluated.

Model structure

A family of personal LPMs

The LPM family combines neural-state tracking with exploration of future neural-field dynamics. Personal versions, explicit horizons, and retained evaluation reports connect the model family to continual digital twins.

Examine the core
Evidence

A reproducible comparison

The public account is qualitative. Dataset identifiers, metrics, baselines, and a reproducible protocol are needed before comparative-performance claims can be assessed.

Read the evidence
Shared infrastructure

A reusable agentic stack

The plan reuses backend data management, training, evaluation, deployment, and monitoring. Frontend agents support users and guide nine proposed use cases; each experience still needs its own evaluation.

Compare the programmes
Market sequence

Match the validation burden

Begin with research collaboration and bounded partner studies, then advance applications as evidence and integration mature.

Explore the market sequence

Current development status

A clear view of the work.

Component Current framing
Digital-twin model prototype Prototype and internal comparison reported by the company; the full benchmark protocol and numerical results are not published here.
LPM family & continual personalization Core model and continual-adaptation directions; model-specific benchmarks and longitudinal adaptation evidence are not published here.
Independent validation Not yet established.
Manuscripts In preparation; ask about current draft availability.
Agentome / backend + frontend agents Twelve planned roles across technical orchestration and user guidance; integrated workflow in development.
Superlearning Proposal; prospective study not yet run.
Restoration Proposal; comparative guidance study not yet run.
Clinical programme Proposal; longitudinal recording collaborators sought.
Clinical deployment None. The site describes research infrastructure.
Commercial plan Ingest, Fit, Serve is a proposed model. This site does not publish customer, revenue, or adoption metrics.

Planned business model

Ingest. Fit. Serve.

The proposed commercial model has three parts: ingest and manage BCI recordings, fit and continually evaluate personal LPMs, and serve model outputs plus agent-supported experiences through partner software. Backend orchestration supports all three; these remain development and pricing concepts.

01 / Ingest

Data platform

Proposed paid data-preparation work: document recording quality, preprocessing, annotations, and session provenance. Subscription terms remain a planning assumption.

02 / Fit

Continual LPM personalization

Proposed training and continual-adaptation engagements that deliver accepted personal model versions and evaluation records. Per-model and enterprise pricing are options to assess with partners.

03 / Serve

Agentome integrations

Proposed interfaces for evaluated model outputs, operational records, and frontend support and guidance. No production API specification, service level, or usage pricing is published here.

Development priorities

Evidence before
application claims.

01

Replicate the twin benchmark

Expand comparisons, investigate failures, and support independent evaluation of the structure-preserving digital twin against conventional state models.

02

Build the agent workflows

Connect data management, continual model updates, evaluation, deployments, and monitoring to frontend support and guided sessions. Evaluate operational errors and usefulness independently.

03

Find the first design partners

Define bounded studies with research, wellbeing, and education partners, with prospective success criteria.

04

Develop the integration layer

Explore data-platform, fitting, and API interfaces while documenting what each endpoint can and cannot support.

Three proposed API directions

A proposed research layer.
Partners retain their applications.

Education

Superlearning API

A proposed integration of personal LPM outputs, Track Designer guidance, and user support for learning, focus, and pacing studies.

Wellbeing

Restoration API

A proposed integration of personal model outputs and frontend guidance for meditation, personalized music, art, sleep, and restoration research.

Research

Clinical research API

A proposed integration of longitudinal twins, perturbation analysis, and investigator-reviewed hypotheses for neurological and mental health research.

Collaborate with Mindverse

Look closely
at the science.

Discuss the prototype status, proposed studies, development costs, market assumptions, and current availability of technical or manuscript materials.

Discuss the company