Batch anomaly detection: check up to 100 expenses, salaries, or metrics against real-world benchmarks in one request.
What it does
AgentUtil Norm's batch endpoint lets AI agents submit up to 100 values at once for statistical anomaly detection against published industry benchmarks sourced from BLS, GSA, and IRS data. Each item in the batch specifies a category (e.g. expense:travel:hotel_nightly, salary:software_engineer), a numeric value, and optional context like unit, region, and company size. The service returns per-item percentile rankings, assessments, and a batch summary indicating how many items were flagged versus normal.
The endpoint is part of the AgentUtil Norm service, which provides baseline reality checks for autonomous AI agents. It answers the question "is this number normal?" via pure data lookup with sub-50ms latency. A companion single-check endpoint (/v1/check) handles one-off queries, while /v1/categories lists the full taxonomy of available benchmark categories.
Pricing for the batch endpoint is $0.001 per item, settled via the x402 protocol using USDC on Base. A free tier of 10 queries per day per IP is available with no signup required. The service publishes an OpenAPI 3.1 spec and agent discovery files (agent-service.json, agent.json) for automated integration.
Capabilities
Use cases
- —An autonomous agent reviewing a batch of employee expense reports to flag outliers before reimbursement
- —Validating salary offers across multiple roles against BLS benchmarks during hiring workflows
- —Screening a set of vendor invoices for anomalous line items before approval
- —Automated financial auditing where many metrics need simultaneous baseline comparison
- —Agent-driven budget analysis checking dozens of cost categories at once
Fit
Best for
- —AI agents that need to validate many numeric values against real-world benchmarks in a single call
- —Expense management workflows requiring bulk anomaly screening
- —Autonomous financial agents performing sanity checks on salaries, costs, or metrics
Not for
- —Custom or proprietary benchmark data — this uses published government/industry sources only
- —Real-time streaming anomaly detection on time-series data
- —Non-numeric or unstructured data validation
Quick start
curl -X POST https://norm.agentutil.net/v1/batch \
-H "Content-Type: application/json" \
-d '{
"items": [
{"category": "expense:travel:hotel_nightly", "value": 450, "unit": "USD", "region": "US"},
{"category": "salary:software_engineer", "value": 250000, "unit": "USD", "region": "US"}
]
}'Example
Request
{
"items": [
{
"unit": "USD",
"value": 450,
"region": "US",
"category": "expense:travel:hotel_nightly"
},
{
"unit": "USD",
"value": 250000,
"region": "US",
"category": "salary:software_engineer",
"company_size": "enterprise"
}
]
}Response
{
"results": [
{
"unit": "USD",
"value": 450,
"summary": "$450/night is at the 88th percentile for US hotel rates.",
"baseline": {
"p25": 120,
"p50": 195,
"p75": 325,
"p90": 480,
"source": "GSA per-diem rates"
},
"category": "expense:travel:hotel_nightly",
"assessment": "high",
"percentile": 88
},
{
"unit": "USD",
"value": 250000,
"summary": "$250,000 is at the 95th percentile for US software engineer salaries.",
"baseline": {
"p25": 95000,
"p50": 135000,
"p75": 185000,
"p90": 230000,
"source": "BLS OES"
},
"category": "salary:software_engineer",
"assessment": "very_high",
"percentile": 95
}
],
"summary": {
"total": 2,
"normal": 1,
"flagged": 1
}
}Endpoint
Quality
Full OpenAPI 3.1 schema with request/response definitions, clear pricing, and a landing page with examples. However, the x402 probe did not return a 402 challenge (returned 400 instead), and no dedicated docs page exists. The example response is inferred from the schema since no live response was captured.
Warnings
- —x402 probe returned HTTP 400 on POST rather than the expected 402 challenge — the endpoint may require a valid request body before issuing a payment challenge, or x402 gating may not be active on this path.
- —No dedicated documentation page found (/docs, /api, /pricing all return 404).
- —Example response is inferred from the OpenAPI schema, not captured from a live call.
Citations
- —Batch endpoint checks up to 100 items against real-world benchmarks (BLS, GSA, IRS data)https://norm.agentutil.net
- —Batch pricing is $0.001 per item via x402 protocol (USDC on Base)https://norm.agentutil.net
- —Free tier: 10 queries/day per IP, no signup requiredhttps://norm.agentutil.net
- —OpenAPI 3.1 spec describes /v1/batch accepting up to 100 items with category, value, unit, region, and company_size fieldshttps://norm.agentutil.net/openapi.json
- —Pure data lookup with sub-50ms latencyhttps://norm.agentutil.net