Getting started

HubSpot is available as a live connector. Jobber is a sandbox integration; Vapi is not currently available as a live connector.

Supported sources use read-only credentials. Records are retained before KPI events are derived. Classify discovered actors as AI agents or humans before scoring; people are never scored.

Re-syncs skip unchanged data.

Upload one row per timestamp. Confirm automatically matched KPI columns and units before writing events. Interrupted uploads can resume, and repeated files skip duplicate records.

Date,Task Success Rate (%),Response Time (s),Customer Satisfaction (%)
2026-08-02,92.4,44,91.8
2026-08-05,93.1,41,92.3

Remove unnecessary personal information first. Original uploaded records are retained.

Submit events against the KPI registry. Unknown metrics and incorrect units are rejected; source record identifiers make retries safe.

# 1. Identify your key (scopes + company binding)
curl https://api.orbscore.org/v1/me \
  -H "Authorization: Bearer $AGENTSCORE_KEY"

# 2. Push KPI events through the validated write path
curl -X POST \
  https://api.orbscore.org/v1/companies/$COMPANY/agents/$AGENT/events \
  -H "Authorization: Bearer $AGENTSCORE_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "source": "my-platform",
    "events": [{
      "kpiName": "task_success_rate",
      "value": 87.5,
      "unit": "percent",
      "recordedAt": "2026-08-01",
      "sourceRecordId": "batch-2026-08-01-tsr"
    }]
  }'

# 3. Recompute and read the scorecard
curl -X POST https://api.orbscore.org/v1/companies/$COMPANY/recompute \
  -H "Authorization: Bearer $AGENTSCORE_KEY"