Simulation and human-machine teaming

Put governed agents inside the mission software you already trust.
Keep the mission in command.

Stark brings agent teams into government software while native policy stays in command. BLACKBODY makes that promise visible through simulation and human-machine teaming, the leading proof for a runtime built to serve wider government missions.

MISSION GRAPH / ILLUSTRATIVE POLICY BOUND
Concurrent agent runtime topologyIllustrative target architecture in which concurrent agents route typed proposals through one governed Stark runtime, designed to connect licensed local models or an authority-approved frontier endpoint with mission tools and evidence.AGENT 01SLOT 01AGENT 02SLOT 02AGENT 03SLOT 03STARK RUNTIMETOOLS · POLICY · EVIDENCELOCAL MODELSDISCONNECTEDFRONTIER ENDPOINTAUTHORITY APPROVEDMISSION TOOLS
ONE GOVERNED CONTROL PLANELOCAL · CONNECTED · HYBRID
Host-authoritativeMission software decides
Model-neutralLocal and approved frontier options
Mission-specificQualified in the named environment

The capability outcome

Governed intelligence. Mission authority intact.

Agents belong inside the mission system. Not in another chat window. BLACKBODY turns that promise into something buyers can see through natural-language intent, autonomous commanders, bounded decisions, and observable consequences inside one live simulation.

Reference evidence · The leading proof

BLACKBODY puts the control model under pressure.

BLACKBODY lets teams examine how Stark handles human-machine teaming and multiple autonomous agents while the simulator retains authority over every outcome.

01 / MODE

Human vs AI

Currently being built

Let an operator express intent in natural language while an autonomous opposing commander responds to the same evolving simulation.

How it works
A person types an objective, an agent translates that intent into a typed fleet-action proposal, and simulator policy accepts, rejects, constrains, or requests clarification before anything changes.
Who decides
The simulator retains state, time, scoring, policy, permissions, action admission, and every consequence.

02 / MODE

AI vs AI

Currently being built

Observe two autonomous commanders adapt to each other in one visible, repeatable mission run.

How it works
Each commander has an isolated identity, session, and tactical view while both use the same governed runtime to submit typed fleet-action proposals.
Who decides
The simulator remains authoritative over state, time, scoring, policy, permissions, action admission, and every consequence.

Evidence state: Mode status is published from the sanitized BLACKBODY evidence record. It changes only when the complete demonstration readiness gate allows publication.

Authority that stays native

Agents propose. Mission software decides.

Stark creates room for useful reasoning while the mission system keeps the powers that define reality.

01 / DECISION

State the mission outcome

A person or autonomous commander works from a bounded view of the mission, the role it has been given, and the decisions available in that moment.

02 / DECISION

Turn intent into a clear proposal

The agent uses approved context and tools to propose a typed action that the mission software can inspect before anything changes.

03 / DECISION

Let native policy decide

The host accepts, rejects, constrains, or requests clarification, then owns the resulting state, timing, scoring, and consequence.

From intent to consequence

Every proposal passes through the mission's rules.

The host opens a bounded decision, an agent proposes a typed action, native policy returns an accepted, rejected, constrained, or clarification outcome, and only the host changes mission state.

  1. 01

    Observe

    Host emits bounded state

  2. 02

    Constrain

    Runtime exposes available authority

  3. 03

    Propose

    Model returns a typed action

  4. 04

    Admit

    Native policy accepts or rejects

  5. 05

    Apply

    Host owns the consequence

  6. 06

    Record

    Facts support audit and replay

Model choice without control drift

Choose the model for the environment. Keep one control model.

Use licensed local models where connectivity is limited, approved frontier models where policy permits, or both. Stark keeps typed tools, permissions, lifecycle, and evidence boundaries consistent around every agent.

Explore deployment patterns
ONE GOVERNED BOUNDARYQualified per environment
Licensed local models
Constrained or disconnected
Approved frontier models
Connected when permitted
Across both choices
Tools, permissions, lifecycle, evidence
MODEL CHOICE DOES NOT CHANGE HOST AUTHORITY

Beyond the proving ground

One governed runtime. Many government missions.

Simulation is the leading proof, not the limit of the runtime. The same governed agent foundation can support cyber operations, technical data, tactical intelligence, acquisition workflows, and government service operations through mission-specific integration.

01

Cyber + mission C2

Coordinate specialist roles around alerts, courses of action, and policy-owned approvals.

02

Technical data

Help teams work across manuals, engineering records, maintenance history, and source provenance.

03

Training + simulation

Explore opposing forces and human-machine teaming while the simulator keeps authority over consequence.

04

Tactical intelligence

Bring bounded analysis and tool use closer to constrained and intermittently connected environments.

05

Acquisition workflows

Move document-heavy analysis through explicit roles, controls, and review gates.

06

Government operations

Add durable agent teams to service and mission-support software already in use.

Scope note: These are candidate applications, not claims of agency adoption or fielding. Each requires mission-specific integration and qualification.

A practical path into existing software

License the runtime, or bring Pinpoint into the integration.

Give your own product team a governed agent foundation, or add Pinpoint engineers as a specialist subcontractor around the mission seam you already own.

01

Runtime path

License Stark

Choose this when
For product and program teams that want to build governed agent capability into software they own.
What we shape together
The Stark Runtime, integration guidance, and a qualification plan shaped around your host, models, tools, and environment.
What stays yours
Own the product experience and integration while keeping model choice, authority, and evidence boundaries explicit.
Discuss Stark licensing
02

Integration path

Build with Pinpoint

Choose this when
For primes and programs that want a specialist team to add governed agents to existing mission, enterprise, simulation, or edge software.
What we shape together
Mission discovery, agent and tool design, adapter engineering, evaluation support, and deployment-specific verification.
What stays yours
Reach a qualified pilot faster without replacing the system of record or building the runtime foundation from scratch.
Scope a mission integration

Start with visible proof

See the agents work.
See the mission stay in control.

Bring an unclassified mission question. We will show the BLACKBODY modes, explain the evidence boundary, and map the same control model to your software.