Skip to main content

AI Endpoint

Ask natural-language questions about any professional's profile and get evidence-backed answers. The response includes structured metadata — answer status, confidence level, and cited sources — so your application can display results intelligently.

Base URL
JWT set Key set No credentials set

Contract Notes

Grounding-first behavior

The service attempts deterministic profile answers first, then falls back to a full-profile LLM path when synthesis is required.

Stable client signals

Use answer_status, answer_mode, confidence_band, and evidence_types to drive UI and automation safely.

AI

1

Ask a natural-language question about a public profile.

Path Parameters

FieldTypeDescription
username string Target username*

Request Body

FieldTypeDescription
question string Question between 3 and 500 characters*

Response

FieldTypeDescription
answer string Grounded answer text*
username string Target username*
answer_status string answered, partially_answered, insufficient_evidence, unsupported, or no_profile_data*
answer_mode string deterministic or llm_full_context*
confidence_band string high, medium, or low user-facing support signal
confidence number Heuristic internal confidence detail
evidence_types array Grounding facets used for the answer*
sources array Compact grounded citations or supporting source rows*

                    
                

Request

{
  "question": "How many years of backend experience does this person have?"
}

Response 200 OK

{
  "answer": "The profile shows about 8 years of backend-oriented experience across Python and API-focused roles.",
  "username": "hershel",
  "answer_mode": "deterministic",
  "answer_status": "answered",
  "confidence_band": "high",
  "confidence": 0.91,
  "evidence_types": ["experience", "skill"],
  "sources": [
    {
      "label": "Work Experience (3)",
      "facet": "experience",
      "snippet": "3 work experience entries in profile"
    }
  ]
}