Evidence-bounded claim verification for RAG and AI-agent workflows.
Start API Access — $49/moAI agents and RAG systems hallucinate. They confidently state false claims as fact. Retrieval-augmented generation helps, but retrieval is not verification — retrieved passages may be irrelevant, outdated, or contradict the claim entirely. AXIMUR closes the gap: it evaluates whether a claim is actually supported by the evidence you provide.
Input: A claim + a grounding source (text, document, or retrieved passages).
Output: A structured verdict — SUPPORTED, UNSUPPORTED, or UNCERTAIN — with confidence score, step-by-step reasoning, and a full audit trail citing the exact evidence spans.
You supply the grounding source. AXIMUR does not search the web. It does not query a knowledge base. It evaluates the claim only against the evidence you provide. This is evidence-bounded verification — not open-web fact-checking.
Security Note: API keys must be kept server-side. Never expose API keys in client-side code, mobile apps, or public repositories.
POST https://api.aximur.com/v1/verify
| Header | Required | Value |
|---|---|---|
Authorization |
Yes | Bearer YOUR_API_KEY |
Content-Type |
Yes | application/json |
{
"claim": "The global average temperature in 2023 was 1.45°C above pre-industrial levels.",
"grounding_source": "According to the WMO State of the Global Climate 2023 report, the global average temperature in 2023 was 1.45°C above the pre-industrial baseline (1850–1900)."
}
{
"verdict": "SUPPORTED",
"confidence": 0.96,
"reasoning": "The claim matches the WMO report's stated figure of 1.45°C above pre-industrial levels (1850–1900 baseline). The grounding source directly supports the specific figure and baseline.",
"audit_trail": [
{
"claim_span": "global average temperature in 2023 was 1.45°C above pre-industrial levels",
"source_span": "global average temperature in 2023 was 1.45°C above the pre-industrial baseline (1850–1900)",
"match_type": "exact_match"
}
]
}
Start API Access — $49/mo
| Verdict | Meaning |
|---|---|
SUPPORTED |
The grounding source directly supports the claim. |
UNSUPPORTED |
The grounding source contradicts the claim or explicitly refutes it. |
UNCERTAIN |
The grounding source is insufficient, ambiguous, or does not address the claim. |
Each verdict includes a confidence score (0.0–1.0) and a detailed reasoning trace with an audit trail linking claim spans to source spans.
curl -X POST https://api.aximur.com/v1/verify \
-H "Authorization: Bearer YOUR_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"claim": "The global average temperature in 2023 was 1.45°C above pre-industrial levels.",
"grounding_source": "According to the WMO State of the Global Climate 2023 report, the global average temperature in 2023 was 1.45°C above the pre-industrial baseline (1850–1900)."
}'
AXIMUR / AXIMUR Security
contact@aximur.com
https://aximur.com