The instrument behind the 30-Day AI Economics Audit. Load your usage exports; receive true AI spend, unallocated cost, cost-per-outcome verdicts, margin forecasts, and a board-grade report — sealed in a verifiable SHA-256 evidence chain. Nothing leaves this browser. No upload, no server, no telemetry.
Expected columns: date, provider, model, workload, team, calls, input_tokens, output_tokens, cost_usd, status · status ∈ ok / retry / failed · empty workload = unallocated spend. Export from OpenAI, Anthropic, Azure AI Foundry, Bedrock or Vertex usage dashboards and map columns to this schema.
The provider invoice is only part of true AI cost. Columns: category, description, monthly_cost_usd — SaaS copilots, vector infrastructure, human review, labeling, compliance, committed minimums.
For each allocated workload, enter monthly outcomes delivered and the defensible value per outcome. The X-Ray converts activity into economics: cost per outcome, value ratio, and a capital verdict. Activity is not value — this is where the difference becomes a number.
Model spend growth against the approved budget. Retry amplification and hidden costs scale with volume; contracted lines do not.
Every stage of this audit — intake fingerprint, ledger, outcome verdicts, forecast, report — is committed as a block whose hash includes the hash of the block before it. Alter any figure and the chain breaks. This is the same architecture as BLACKBOX and THE RECORD, pointed at money.
This instrument is delivered as part of a governed engagement: expenditure mapping, outcome attribution workshops, margin-exposure analysis, portfolio funding recommendations, and an executive report your board can rely on — with the evidence pack to prove it.
richardleclezio.com · Leclezio Consulting Corporation · The Studio