Describe the bounded situation you want to model. After I produce a draft, tell me what to change and I will generate the next saved revision.
Choose and play
Service desk
A bounded incident workflow with an authored service-operations center, a remote customer site, and one login incident.
Model and simulation assumptions
Loading…
What this scenario represents
Assumptions
Known omissions
Fidelity questions
Simulation map
Choose a view, then select a node or edge for its retained state. Spatial topology shows authored places and physical links; configured interaction pathways show the scenario relationships through which information or action may travel; the realized causal graph shows what actually occurred.
The graph will change only when the runtime retains public evidence.
What happened
Start with the situation, then follow the decisions and events that changed the outcome. Detailed mode explains what each action changed; exact technical evidence remains available on demand.
How coordination changed
Recorded by the simulator
Model interpretation
Decision snapshot
Interpretations to review
Direction means change relative to this run’s starting state. These judgments describe only the supplied synthetic evidence; they do not establish truth, intent, or an organization-level mind.
All recorded measures
Limits and model-call provenance
What this run lets us inspect
These are separate theoretical views over the same retained trajectory. They expose candidate observations and their exact evidence; they are not validated estimates of hidden organizational state.
Waltzman-informed view
Decision-environment instrument
Candidate trust-, risk-, and coordination-related observations from one run, with chronology and limits kept explicit.
Levin-informed view
Collective competence
What the configured group boundary accomplished relative to its goal, and which capacities were observed or remain untested.
Follow a participant or view the team as a whole
Choose a person to follow their decisions, a process or source to see what it produced, or a group to inspect the retained outside inputs, responses, and result.
Advanced Run settings, exact mechanisms, and retained evidence
Inspect one exact moment Synchronize the map with the retained event trace
Choose a moment
The map, account, and inspector stay synchronized with this moment. Simulated time may advance unevenly because each process follows its own cadence.
Raw retained evidence
Run history
Open an earlier run for inspection or move it to recoverable server trash. Each entry includes its exact run ID.
Collective capability stress test
Compare five matched conditions
See whether the same five-person partnership still reaches a valid decision after replacing one member, interrupting a route, removing verification feedback, or introducing a legitimate external risk.
This reviewed comparison uses fixed reference behavior, makes no model calls, and tests the simulation instrument—not a real organization.
Coordination dynamics experiment
Compare pressure, adaptation, and validation
Run baseline, fixed heterogeneous pressure, feedback-adaptive pressure, and adaptive pressure with authoritative validation as one retained eight-run experiment.
This synthetic reference experiment makes no model calls. Its directional contrasts apply only to the retained batch and step down to ordinary exact runs.
Live theory probe
Run the same conditions with LLM-modeled participants
Select Coordination decision in the simulation controls, then choose a condition labeled Experiment. The existing model, spend authorization, progress, pause/stop, and retained-run controls all apply.
Run one condition at a time. A single trajectory is inspectable mechanism evidence, not an effect estimate; use the scripted reference matrix below as the lifecycle control.No retained runs yet.
How to read a cybernetic simulation
This is a bounded causal model, not a claim that a real organization is a single mind. People and exact processes act; records, credentials, and mechanisms constrain what can happen; an organization is an analytical composite over those retained components.
Choose and play
Select a scenario and condition, then run it. Live runs let LLM-modeled people act from their retained memory, position, observations, and interfaces. Reference runs use fixed zero-cost policies.
Read the map
Spatial topology shows authored places, occupants, and physical links. Configured interaction pathways show relationships that may support information or action when exact mechanisms permit them; they do not grant authority. Realized causal graph shows what happened in the selected run.
Change analytical scale
The graph’s analytical-scale control collapses or expands an execution-inert composite such as the service desk. It changes what is summarized, not what exists or acts in the simulation.
Follow the account
The concise story follows the people and other active participants. Detailed mode explains what their actions changed; Advanced lets you inspect individual retained events, mechanisms, and state.