Governed runtime
Coordinate models and agents with evidence of what ran, where, under which policy, and with what result.
Portfolio / Compute
Infrastructure
Compacitas
A verifiable governed runtime for model execution and optimization.
The company
A verifiable governed runtime for model execution and optimization.
Compacitas is a Verifiable Governed Runtime Environment - the civilization runtime for model governance, scheduling, and optimization. It governs models and agents during execution, schedules mission workloads, and produces verifiable evidence of what ran, where, under which policy, and with what result. Compact secure clusters bring the runtime into mission environments; Multiplex increases usable GPU capacity, and the Grid federates authorized capacity across facilities. A civilization node is a software deployment that can run on customer infrastructure or be packaged with compute hardware. Morphitas is a separate experience engine connected through an adapter; the runtime remains responsible for civilization state, authorized execution, and node coordination.
The problem it addresses
Model execution, workload placement, and resource optimization need governance that remains visible throughout operation, including inside secure mission environments.
Capabilities
What Compacitas brings.
Multiplex
Increase usable GPU capacity through scheduling and optimization connected to the workload's operating requirements.
Grid
Federate authorized capacity across participating facilities while compact clusters bring the runtime into existing environments.
Civilization nodes
Package the runtime as software or with a compute node. Connect Morphitas through an adapter while retaining independent execution authority and source records.
Illustrative application
Compacitas in context.
Models and agents execute within a governed runtime. Scheduling and optimization allocate authorized capacity; execution evidence records what ran, where, under which policy, and with what result.
A practical starting point
- Describe the mission workload, model requirements, available resources, and execution policy.
- Schedule authorized work and coordinate runtime behavior within the permitted environment.
- Review execution evidence and resource use to inform optimization and the next operating decision.
What to evaluate
- Conformance with execution and placement rules
- Usable capacity within workload constraints
- Completeness of runtime and scheduling evidence
The intended outcome
A verifiable civilization runtime connecting model governance, resource optimization, Multiplex, and authorized federation.
This application describes product direction and an evaluation approach, not a completed customer deployment or independently validated performance.
Inside the company
Products and research.
1 products & product families
View company products
10 research projects
View company research
MultiplexCompacitas · Prototype
Multiplex
Scheduling many persistent agent identities across bounded model residency and physical compute.
AletheonCompacitas · Protocol in development
Aletheon
A controlled emergence ecology for testing persistent identity, history, diverse substrates, and governance needs.
Verifiable Civilizational LearningCompacitas · Proposed study
Verifiable Civilizational Learning
Testing whether verified experience can accumulate outside model weights and survive complete agent and model turnover.
A System of Heredity for IntelligenceCompacitas · Research design
A System of Heredity for Intelligence
Checked inheritance of commitments and capability across generations of neural and non-neural computational systems.
Precedent EngineCompacitas · Research framework
Precedent Engine
Turning reviewed outcomes into scoped, challengeable lessons and executable controls.
Neural InterchangeCompacitas · Working draft
Neural Interchange
Scoped exchange of identity, capabilities, results, and evidence across independently governed AI systems.
ChronicleCompacitas · Prototype
Chronicle
A verifiable chronology of decisions, execution, learning, and change.
World State and JEPACompacitas · Research design
World State and JEPA
Grounded world models that distinguish observations, inferred state, predictions, actions, and outcomes.
Federated Cancer Hypothesis GridCompacitas · Feasibility study
Federated Cancer Hypothesis Grid
Persistent, auditable hypothesis testing across pediatric and young-adult AML datasets.
Genesis Governance and AssuranceCompacitas · Working draft
Genesis Governance and Assurance
Executable assurance requirements for identity, authority, evidence, privacy, and recovery in agent systems.
Connected ecosystem
Where the capabilities meet.
Explore the complementary companies that connect Compacitas to the wider product journey.
ComputeCapacitas
A governed compute civilization across federated multi-compute cluster resources.
AIArchetypal AI
Archetypal AI Research for emergent behavior and alignment; Guarded for operational governance, audit, compliance, and risk.
CyberCybersecutitus
A governed AI civilization for secure product assessment, cybersecurity validation, and authorized cyber missions.
AIMorphitas
The independent experience engine for personal and enterprise work, bringing connected information, decisions, and actions into chat.
Related perspectives
Research note / AI
Beyond the harness: building with an AI civilization
A model connected to tools is not the same thing as a coordinated environment for product development.
Read perspective
Research note / Compute
Federation without surrendering local control
A compute civilization must respect the ownership and authorization of participating resources.
Read perspective