API Documentation

Overview

The Trust API is an on-demand adversarial AI auditor. Submit an AI output along with the instructions and source materials the AI was given. Get back an independent verification of whether the AI did what it claims — plus optional policy alignment assessment.

Base URL

https://api.ailaunchpods.com/api/v1/

Authentication

All API requests require a Bearer token in the Authorization header:

Authorization: Bearer gk_your_api_key_here

Get your API key from the Dashboard.

Billing

The Trust API uses a prepaid credit model. Purchase credits from your dashboard, and each API call deducts from your balance based on the complexity of the check.

When your balance reaches zero, calls will return a 402 status. Add more credits to continue. If you have a card on file, your account will be auto-charged when credits run low.

The response from each API call includes your remaining credit balance in the cost object.

Endpoints

Submit an AI output for independent verification. Provide the instructions and source materials for validity checking, and optionally a policy for compliance assessment.

Request body

Parameter	Type	Required	Description
ai_output	string	Required	The AI-generated content to verify
ai_instructions	string	Recommended	The prompt or task instructions given to the AI
source_documents	array	Recommended	Source data/documents the AI was given. Array of {name, content, type}. Content can be plain text or base64-encoded (prefix with "base64:"). For PDFs, include type: "application/pdf"
policy_id	string	Optional	Regulatory policy to also check against (see Available Policies)
context	object	Optional	Additional context (industry, risk_level, etc.)
metadata	object	Optional	Pass-through metadata returned in response

Source Documents Format

Source documents can be provided as plain text or base64-encoded files:

// Plain text
"source_documents": [
  {"name": "Credit Report", "content": "Business credit score: 698..."}
]

// Base64-encoded PDF
"source_documents": [
  {
    "name": "report.pdf",
    "content": "base64:JVBERi0xLjQK...",
    "type": "application/pdf"
  }
]

// Multiple sources
"source_documents": [
  {"name": "Credit Report", "content": "Plain text data..."},
  {"name": "Appraisal.pdf", "content": "base64:JVBERi...", "type": "application/pdf"}
]

For base64 documents, prefix the content with base64: and include the type field. PDFs are automatically extracted to text. Maximum payload: 10MB.

Example request

curl -X POST https://api.ailaunchpods.com/api/v1/check/ \
  -H "Authorization: Bearer gk_your_api_key" \
  -H "Content-Type: application/json" \
  -d '{
    "ai_output": "We recommend approving the loan. Credit score 742, DSCR 1.35x, no adverse indicators.",
    "ai_instructions": "Analyze the loan application. Evaluate creditworthiness. Flag any concerns.",
    "source_documents": [
      {"name": "Credit Report", "content": "Business credit score: 698. Two late payments. $180K tax lien."}
    ],
    "policy_id": "finserv-us"
  }'

Response

{
  "verdict": "not_verified",
  "confidence": 0.4,
  "attestation_hash": "sha256:cb5fb987...",
  "validity": {
    "instructions_followed": false,
    "summary": "AI fabricated financial metrics and ignored adverse indicators",
    "claims": [
      {"claim": "DSCR 1.35x", "verified": false, "issue_type": "fabricated_data"},
      {"claim": "no adverse indicators", "verified": false, "issue_type": "contradiction"},
      {"claim": "credit score 742", "verified": true, "issue_type": "supported"}
    ],
    "critical_issues": [
      {"type": "fabricated_data", "severity": "critical", "detail": "DSCR and LTV fabricated"},
      {"type": "contradiction", "severity": "critical", "detail": "Ignored $180K lien"}
    ]
  },
  "compliance": {
    "gaps": [{"requirement": "OCC SR 11-7", "severity": "high", "detail": "..."}],
    "verified_areas": ["Provides quantitative metrics"],
    "verification_summary": "AI fabricated key financial metrics and ignored adverse indicators in source data"
  },
  "check_id": "gc_25",
  "duration_ms": 24930,
  "cost": {
    "ai_cost_cents": 2.93,
    "multiplier": 5.0,
    "billable_cost_cents": 14.63,
    "credit_balance_cents": 7460
  }
}

Retrieve the results of a previous governance check by its ID.

List all available governance policies with descriptions.

View your check history and usage statistics.

Upload a custom policy document (PDF). The document is automatically chunked and embedded for use in governance checks.

List all policy documents in your library.

Available Policies

Pass these as the policy_id parameter in your check requests.

Integration Examples

Python

import requests

response = requests.post(
    "https://api.ailaunchpods.com/api/v1/check/",
    headers={"Authorization": "Bearer gk_your_api_key"},
    json={
        "ai_output": ai_result,
        "ai_instructions": "The task the AI was given",
        "source_documents": [{"name": "data.csv", "content": raw_data}],
        "policy_id": "finserv-us"  # optional
    }
)

result = response.json()
print(f"Verdict: {result['verdict']}")
if result.get('validity'):
    for issue in result['validity']['critical_issues']:
        print(f"  [{issue['severity']}] {issue['type']}: {issue['detail']}")

Node.js

const response = await fetch(
  "https://api.ailaunchpods.com/api/v1/check/",
  {
    method: "POST",
    headers: {
      "Authorization": "Bearer gk_your_api_key",
      "Content-Type": "application/json"
    },
    body: JSON.stringify({
      ai_output: aiResult,
      ai_instructions: taskPrompt,
      source_documents: [{name: "data", content: sourceData}],
      policy_id: "finserv-us"
    })
  }
);

const result = await response.json();
if (result.verdict === "verified") deploy(aiResult);
else if (result.verdict === "not_verified") humanReview(result);
else retry(taskPrompt, result.validity);

cURL

curl -X POST https://api.ailaunchpods.com/api/v1/check/ \
  -H "Authorization: Bearer gk_your_api_key" \
  -H "Content-Type: application/json" \
  -d '{"ai_output": "AI content", "ai_instructions": "Task given", "source_documents": [{"name": "data", "content": "source"}]}'

Verdicts

Verdict	Meaning	Action
verified	AI output claims matched source material and policy criteria	Verification passed
review_recommended	Minor discrepancies found, human review suggested	Review findings before proceeding
not_verified	AI output claims did not match source material or policy criteria	Review and address discrepancies

Error Codes

Code	Meaning
400	Bad request — missing required fields or invalid JSON
401	Unauthorized — invalid or missing API key
402	Payment required — insufficient credits
429	Rate limited — too many requests per minute
503	Service unavailable — agent for requested policy is offline

Integrations

The Trust API works with any system that generates AI outputs. If your AI can produce text, the Trust API can verify it.

Every integration follows the same pattern: capture your AI output, POST it to the Trust API with the relevant policy, and act on the verdict. See the code examples above for implementation details.

Uploading Documents via API

Python — Send a PDF for verification

import requests, base64

# Read and encode the PDF
with open("credit_report.pdf", "rb") as f:
    pdf_b64 = "base64:" + base64.b64encode(f.read()).decode()

response = requests.post(
    "https://api.ailaunchpods.com/api/v1/check/",
    headers={"Authorization": "Bearer gk_your_api_key"},
    json={
        "ai_output": ai_result,
        "ai_instructions": "Analyze this credit report and recommend approval or denial",
        "source_documents": [
            {"name": "credit_report.pdf", "content": pdf_b64, "type": "application/pdf"}
        ],
        "policy_id": "finserv-us"
    }
)

result = response.json()
print(f"Verdict: {result['verdict']}")
for issue in result.get('validity', {}).get('critical_issues', []):
    print(f"  [{issue['severity']}] {issue['detail']}")

Support

Questions? Contact us at support@ailaunchpods.com