Audit AI workflow trace value integrity

What it's for

Gives leaders a trustworthy unit of AI economics: one workflow request, every loop and tool call, its full cost, and one real outcome—not disconnected token and activity totals.

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.

{
  "business_outcomes": [
    {
      "business_loss": 2,
      "evidence_verified": true,
      "execution_id": "execution-a",
      "gross_business_value": 10,
      "id": "outcome-a",
      "outcome_matured": true,
      "unique_attribution_verified": true
    }
  ],
  "step_events": [
    {
      "attempt_number": 1,
      "direct_cost": 1,
      "ended_at_ms": 20,
      "evidence_verified": true,
      "execution_id": "execution-a",
      "id": "plan-event",
      "input_units": 100,
      "output_units": 20,
      "parent_step_event_id": null,
      "satisfied_control_ids": [
        "approval"
      ],
      "sequence_number": 0,
      "started_at_ms": 0,
      "step_kind": "model",
      "succeeded": true,
      "workflow_step_id": "plan"
    },
    {
      "attempt_number": 1,
      "direct_cost": 2,
      "ended_at_ms": 100,
      "evidence_verified": true,
      "execution_id": "execution-a",
      "id": "answer-event",
      "input_units": 200,
      "output_units": 50,
      "parent_step_event_id": "plan-event",
      "satisfied_control_ids": [],
      "sequence_number": 1,

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": [
    "Execution, step-event, control and business-outcome records share one immutable tenant, workflow version, clock and value/cost perimeter; missing events and unfinished executions remain visible.",
    "Parent links encode trace lineage rather than causal contribution. Direct step cost is invoice-compatible and business outcome attribution is independently governed and unique.",
    "A passing trace establishes observability and accounting integrity, not causal incremental value, workflow safety, model/tool quality, or authorization to automate.",
    "Outputs are aggregate workflow evidence and never an assessment of a provider, team or person."
  ],
  "configuration": {
    "maximum_relative_reconciliation_error": 0.005,
    "minimum_evidence_coverage": 0.95,
    "unfinished_rule": "in_progress_executions_are_retained_and_cannot_have_mature_outcomes",
    "value_rule": "gross_business_value_minus_business_loss_minus_reconciled_step_cost"
  },
  "decision": "ai_workflow_trace_value_integrity_supported",
  "execution_diagnostics": [
    {
      "execution_id": "execution-a",
      "failed_gates": [],
      "observed_trace_latency_ms": 100,
      "reconciled_cost": 3,
      "reconciled_net_business_value": 5,
      "relative_cost_error": 0,
      "relative_latency_error": 0,
      "reported_cost": 3,
      "reported_latency_ms": 100,
      "reported_net_business_value": 5,
      "retry_event_count": 0,
      "step_event_count": 2,
      "supported": true,
      "terminal_status": "succeeded",
      "workflow_class_id": "support-resolution",
      "workflow_version": "v2"
    }
  ],
  "failed_gates": [],
  "method": "end_to_end_ai_workflow_trace_cost_value_lineage_audit_v1",
  "summary": {
    "evidence_coverage": 1,
    "execution_count": 1,
    "in_progress_execution_count": 0,
    "supported_execution_count": 1,
    "unsupported_execution_count": 0,
    "workflow_version_count": 1
  },

How it works

Statistical audit & measurement — Check whether a number is fit to decide on: coverage, timing, reconciliation, and the gaps a dashboard hides.

1. 1 Join each immutable workflow-version execution to all step attempts, parent events, satisfied controls and its independently governed mature business outcome, retaining in-progress executions and missing rows.
2. 2 Validate one rooted trace, parent-before-child order, consecutive attempts with no retry after success, required steps/controls, execution-time containment and terminal/outcome cardinality.
3. 3 Reconcile step cost, observed wall-clock latency and gross value minus loss minus cost to reported execution totals; aggregate only after every lineage and evidence failure remains visible.

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": "audit every ai workflow execution from" }
  → finds "audit_ai_workflow_trace_value_integrity"

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

gitrevio_capability_run
  { "capability_id": "audit_ai_workflow_trace_value_integrity", "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 AI cost, routing & return →