AI LAB technical archive
API Reference
Integrate Co-one directly into your CI/CD pipeline or production inference loop.
Endpoints
Evaluate Grounding
Evaluate a single response against a given context.
// Evaluate a response against context
POST /v1/evaluate/grounding
{
"prompt": "What is our bank's limit for instant transfers?",
"response": "The limit is $5,000 for standard accounts.",
"context": [
"Instant transfer limits: Standard ($5k), Premier ($25k)."
],
"metrics": ["grounding", "reliability"]
}Batch File Upload
Perform large-scale regression tests by uploading CSV or JSONL files. Ideal for enterprise model auditing and historical dataset grounding checks.
Field
query
REQUIRED
The user input or prompt string to be evaluated.
Field
output
REQUIRED
The AI agent's generated response to the query.
Field
context
OPTIONAL
The source documents or facts for grounding check.
// Batch evaluation via file upload
POST /v1/evaluate/batch
Content-Type: multipart/form-data
curl -X POST https://api.co-one.co/v1/evaluate/batch \ -H "Authorization: Bearer YOUR_API_KEY" \ -F "file=@/path/to/evaluations.csv" \ -F "project_id=proj_29x7" \ -F "metrics=grounding,reliability,safety"
// Sample Response Output
{
"id": "eval_8k2n9m1",
"score": 0.98,
"reliability": 0.94,
"safe": true,
"violations": [],
"metadata": {
"model_version": "co-one-reliable-v4",
"latency_ms": 142
}
}