Forecast knowledge continuity semimarkov

What it's for

Turns a static bus-factor snapshot into a validated forward view of which critical systems may become orphaned, how long recovery could take, and what the economic tail means for delivery and diligence.

What you give it

Inputs split into evidence read from your connected systems, calibration your team owns, and numerical controls that affect precision but never the result's meaning.

{
  "current_units": [
    {
      "criticality_value": 10,
      "current_state": "concentrated",
      "delay_cost_per_period": 5,
      "evidence_verified": true,
      "id": "payments",
      "recovery_cost": 25,
      "state_age_periods": 1,
      "unit_class_id": "service"
    }
  ],
  "historical_periods": [
    {
      "evidence_verified": true,
      "id": "knowledge-0-0",
      "next_state": "resilient",
      "outcome_matured": true,
      "period": 0,
      "state": "resilient",
      "state_age_periods": 0,
      "unit_class_id": "service",
      "unit_ref": "unit-0"
    },
    {
      "evidence_verified": true,
      "id": "knowledge-0-1",
      "next_state": "concentrated",
      "outcome_matured": true,
      "period": 1,
      "state": "resilient",
      "state_age_periods": 1,
      "unit_class_id": "service",
      "unit_ref": "unit-0"
    },
    {
      "evidence_verified": true,
      "id": "knowledge-0-2",
      "next_state": "orphaned",
      "outcome_matured": true,
      "period": 2,
      "state": "concentrated",
      "state_age_periods": 0,

What you get back

This is the actual output of running the example above — computed by the same function the platform calls, not an illustration.

{
  "assumptions": {
    "age_buckets": [
      "new",
      "established",
      "persistent"
    ],
    "class_shrinkage": "Beta leave hazards and Dirichlet destinations shrink to the company-wide state model",
    "common_scenario_per_draw": true,
    "individual_departure_prediction_permitted": false,
    "simulation_draws": 200,
    "states": [
      "resilient",
      "concentrated",
      "orphaned",
      "recovering"
    ]
  },
  "decision": "accepted",
  "knowledge_unit_forecasts": [
    {
      "expected_loss": 46.225,
      "expected_unavailable_periods": 3.295,
      "knowledge_unit_id": "payments",
      "p90_loss": 67.5,
      "probability_orphaned": 1,
      "supported": true,
      "unit_class_id": "service"
    }
  ],
  "limitations": [
    "The estimand is continuity of a technical knowledge unit, not the probability that a named person leaves.",
    "Scenario multipliers are governed stresses, not learned causal effects.",
    "State definitions and period cadence must remain stable across training, validation, and forecast inputs."
  ],
  "method": "hierarchical_bayesian_semi_markov_knowledge_continuity_v1",
  "portfolio_forecast": {
    "conditional_value_at_risk": 67.5,
    "expected_loss": 46.225,
    "horizon_periods": 6,
    "tail_probability": 0.1,
    "value_at_risk": 67.5
  },
  "validation": {

How it works

Sequential Bayesian & bandits — Learn while deciding — update beliefs as evidence arrives and choose where the next unit of effort is worth spending.

1. 1 Build a complete zero-inclusive mature unit-period panel over stable resilient, concentrated, orphaned and recovering definitions; reserve the latest whole periods before fitting anything.
2. 2 Fit class-shrunk Beta state-exit hazards by duration bucket and Dirichlet next-state distributions, then compare multiclass log loss and leave-probability calibration with a persistence baseline on the untouched holdout.
3. 3 For each current knowledge unit, sample posterior transition dynamics under one coherent operating scenario per draw and price orphaned/recovering periods plus recovery events into unit and portfolio loss, VaR and CVaR; downgrade to diagnostic-only on weak transitions, poor validation, unverified evidence or unseen classes.

Before you trust it

Every tool in the catalog ships with the conditions under which its answer is meaningful — and the conditions under which it should abstain instead of guessing.

Calibrating it to your org

Calibration workflow

1. 1 Define the management decision, target outcome, aggregate unit, privacy boundary, cadence, and prediction/intervention horizon for this organization.
2. 2 Build a tenant-scoped historical cohort using only information available before each prediction or decision; preserve zero periods, censoring, assignment probabilities, and unresolved outcomes when the method requires them.
3. 3 Estimate statistical parameters on training history, but obtain costs, utilities, risk tolerance, practical-effect thresholds, capacity, and policy constraints from accountable decision owners.
4. 4 Validate on later time windows or held-out aggregate units at the deployment grain, against a simple baseline and the function-specific validation strategy.
5. 5 Deploy only if the returned decision clears evidence, overlap, calibration, robustness, and guardrail checks; warning, unsupported, schema-gap, and fallback decisions are abstentions.
6. 6 Monitor realized outcomes, data drift, coverage, and decision regret; recalibrate at a governed cadence or after a detected regime/definition change, never merely because a stakeholder dislikes the result.

Call it from your AI

You don't wire up 388 tools in your MCP client. The GitRevio MCP server exposes 18 tools, three of which let an agent search the catalog, read a tool's schema, and run it — so the assistant finds this one on its own.

gitrevio_capabilities_search
  { "q": "forecast critical codeknowledge continuity with a" }
  → finds "forecast_knowledge_continuity_semimarkov"

gitrevio_capability_describe
  { "capability_id": "forecast_knowledge_continuity_semimarkov" }
  → returns the input schema and agent guidance shown on this page

gitrevio_capability_run
  { "capability_id": "forecast_knowledge_continuity_semimarkov", "arguments": { ... } }
  → returns the result shown above

Works in Claude Desktop, Claude Code, Cursor, Cline, Continue.dev, Goose and Aider. See the MCP server.

Related tools

See every tool in People, retention & knowledge →