Allocate restless bandit interventions

What it's for

Extends single-context LinUCB recommendations into portfolio-wide dynamic allocation when unattended teams or projects continue to evolve.

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.

{
  "evaluation_horizon": 12,
  "intervention_capacity": 1,
  "seed": 4,
  "simulations": 200,
  "units": [
    {
      "active_transition_matrix": [
        [
          0.1,
          0.9
        ],
        [
          0.05,
          0.95
        ]
      ],
      "current_state": 0,
      "id": "payments",
      "intervention_cost": 1,
      "passive_transition_matrix": [
        [
          0.9,
          0.1
        ],
        [
          0.3,
          0.7
        ]
      ],
      "state_values": [
        0,
        10
      ]
    },
    {
      "active_transition_matrix": [
        [
          0.1,
          0.9
        ],
        [
          0.05,
          0.95

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": [
    "Each unit follows the supplied stationary controlled Markov transitions, with state value and intervention cost measured on one additive utility scale.",
    "Whittle allocation is justified only when each relaxed single-unit problem is indexable; otherwise the function withholds a deployment recommendation.",
    "Units evolve even when passive, intervention effects occur through future-state transitions, and at most the declared capacity can be active in a period.",
    "This allocates reversible team- or project-level support under a governed model; it must not be used as an automated employment or individual-surveillance decision."
  ],
  "configuration": {
    "all_units_indexable": true,
    "detail_rows_returned": 3,
    "detail_rows_truncated": 0,
    "discount_factor": 0.95,
    "evaluation_horizon": 12,
    "intervention_capacity": 1,
    "selected_now_count": 1,
    "simulations": 200,
    "unique_unit_models": 1,
    "units": 3
  },
  "decision": "deploy_whittle_allocation",
  "intervene_now": [
    "data"
  ],
  "method": "discounted_whittle_index_restless_bandit_v1",
  "portfolio_evaluation": {
    "expected_discounted_value": 174.4525,
    "incremental_expected_discounted_value": 99.309,
    "incremental_value_interval": {
      "high": 157.3593,
      "low": 45.6107
    },
    "no_intervention_expected_discounted_value": 75.1435,
    "probability_beats_no_intervention": 1
  },
  "unit_indices": [
    {
      "current_state": 0,
      "current_state_value": 0,
      "indexable": true,
      "indices_by_state": [
        16.6744,
        1.4934
      ],
      "intervene_now": true,

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 Allocate scarce recurring interventions across evolving Markov units with Whittle indices, explicit indexability checks, and paired policy simulation.
2. 2 Evaluate the method-specific diagnostics and gates returned by the function, then abstain unless the declared decision clears them under locally governed thresholds.

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": "allocate scarce recurring interventions across evolving" }
  → finds "allocate_restless_bandit_interventions"

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

gitrevio_capability_run
  { "capability_id": "allocate_restless_bandit_interventions", "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 Risk, tails & resilience →