Business Twin

A twin from the records you already keep

Every Action in One3 is logged. Business Twin, planned for W5 and requiring Forecast & Simulation and the Decision Engine, uses that event log to model how the business actually works, then projects what happens next if nothing changes, or if something does.

What it is made of

A twin model combines metric trees, driver models, forecasts and process models, versioned per tenant. Process discovery mines the action log and activity history into process maps, variants, cycle times and bottlenecks. Core KPIs are projected continuously with intervals and refreshed as events arrive.

Maturity you can see

Each KPI and the twin as a whole carry a maturity level from L0 to L5, with the evidence behind each level. Levels are computed from evidence records, never set by hand. The validation protocol is enforced: backtests, naive benchmarks, stress tests, an assumption register, drift monitoring and human sign-off. A twin that stops passing is downgraded automatically.

Guardrails on what it says

"Will not claim" guardrails are enforced through output validation, so the twin does not present projections as certainties. Exploratory agent-based simulation is labelled "exploratory, not predictive".

Intervention simulation

Intervention simulations attach to Decision Engine options so leaders can compare consequences before committing. Reaching L4 requires comparing simulations with real experiment results for a tenant-set number of interventions. System dynamics models cover feedback loops such as hiring, capacity, quality and churn, and discrete-event simulation uses mined process timings. A continuous autonomous twin running tactical loops within autonomy budgets, with governance approval, is a W6 goal.