[ evidence plane · cost per outcome ] plan → authorize → execute → prove → optimize
What the outcome cost — not what each call cost.
A transaction ledger tells you a call cost $0.85. This tells you a competitor report cost $7.19 across 16 calls, $0.85 of which was duplicate spend the controls prevented. The unit is the task — with the spend traced to a signed mandate and an evidence receipt.
$23.19total governed spend
$5.70cost per successful task
$2.30duplicate spend prevented
$1.05substitution savings
100%tasks within budget
$7.80failed-task waste
Cost per outcome · by taskevery row links to its provenance receipt
taskagentcallsspendoutcome
Competitor analysiscompetitor-analysis-8492research-agent-1716$7.19acceptedreceipt Deploy checkout servicedeploy-checkout-svcdeploy-agent-39$4.20acceptedreceipt Lead enrichment batchlead-enrich-batchlead-enrich-agent142$11.80+$1.40 wastepartialreceipt EOD market scanmarket-scan-eodmarket-scan-agent38$0.00+$6.40 wasterejectedreceipt The EOD market scan spent $6.40 but its data failed freshness verification — so it shows as failed-task waste and the provider was auto-refunded. That's the difference between cost-per-call and cost-per-outcome. Every price here is pinned to a verified, versioned offer.
Spend by providerreliability is contextual, per task-domain
market-data.acme$8.40ledgerlite.io$5.10peoplegraph.io$11.80edge-runner.dev$4.20tickwide.exchange$0.00 $6.40 recovered via SLA refunds this period