Organization Building
Is your org's structure sound?
Shape a whole organization — its size, how it's wired, and who has to sign off — then watch how disorder, decision speed and coordination cost play out over 500 simulated runs.
Your setup
Start from
~1 second · in your browser
Tune your org · optional
Changes apply on the next run.
Who has to sign off Approval stack optional
Build the chain of sign-offs a decision clears — 1 to 5 seats, each a person, committee, or AI agent. Tap a seat to set it. This is what sets your decision-latency number.
Results are out of date — run again to refresh the charts
starting engine…
model —
Before / after
Same org — structure decides
Structural Health 7 your org
Disorder
How chaotic things get
Disorder · entropy, 0–100Composite disorder index (0–100), p50 line inside the p10–p90 Monte Carlo band.
Decisions
How fast decisions get made
Decision velocity · 0–100A 0–100 speedometer for decisions: layers add friction, queues drag it lower.
Coordination
How tangled coordination gets
Communication load · open channelsOpen channels inside and between teams; labeled vertical lines mark cognitive band crossings.
Counterfactual
What the AI actually changed
AI injection delta · with-AI minus withoutPointwise difference, AI minus no-AI, on this exact org.
Coordination
Time in meetings, not the work
Meeting overhead · % of capacityShare of each step's capacity spent coordinating instead of working.
Multi-level health
Do the teams feel it too?
Multi-level health · org vs teamsTeam cohesion vs org-level health: divergence means sound teams inside a strained org.
Reproduce
Share this run
The link embeds the seed, the config and the full resolved coefficient set — whoever opens it replays this exact run, byte for byte, in their own browser.
Take it with you
Result card
A 1200×628 image of this run — the decision-latency headline, the throughput-and-entropy story and the research-framing line, ready to post. Generated entirely in your browser; nothing is uploaded.
Under the hood
Every coefficient the engine uses
Extracted straight from MODEL.md with its research anchor, evidence tier and limitation. A CI drift gate keeps this drawer identical to the shipped model — nothing here is hand-copied.
Team Building
Will this team's makeup work?
Compose one team — who does the work, humans and AI, and who reviews it — then see how much gets done, the quality it ships at, how the team holds together, and where essential work goes uncovered.
Your team
Start from
~1 second · in your browser
Results are out of date — run again to refresh the charts
Ready — 500 simulations, in your browser.
model —
Coverage
Is every essential job covered?
Function coverage · build, review, resolveWhether each essential job — building, reviewing, resolving ambiguity — has enough qualified attention on it.
Output
How much work gets done
Throughput · items per stepCompleted work per step — the median line inside the 10th–90th percentile band.
Quality
The quality of what ships
Quality distribution · completed tasksHow the quality of completed work is spread across the run — taller bars on the right mean more high-quality output.
Coordination
Time lost coordinating
Coordination tax · % of capacityThe share of the team's time spent coordinating instead of doing the work.
Trust
How well the team holds together
Cohesion · 0–100Team trust over time — the shared context and willingness to flag problems early that holds a team together.
Fragility
How often work breaks
Brittleness · cumulative breakagesRunning total of breakages when work hits something novel or high-stakes an AI can't handle alone; whether one owns recovery decides if a break spreads.
The recovery cost — how long routine decisions take to clear — surfaces here after a run.
Multi-level health
The team vs its wider org
Multi-level health · team vs orgTeam cohesion against the org-level health it implies: divergence means a sound team inside a strained structure, or the reverse.
Review
Is review keeping up?
Review queue · items waitingWork waiting on review each step. When a fixed reviewer can't keep pace with fast execution, the queue climbs and the wait grows.
Take it with you
Team result card
A 1200×628 image of this team run — the coverage headline, what shipped and at what quality, and the research-framing line, ready to post. Generated entirely in your browser; nothing is uploaded. (Team runs travel as a card — the exact-replay link is Organization-only.)