Reconciliation

Resolve raw entity strings — names, URLs, partial mentions — to a stable canonical Entidex ID. Use this before any other endpoint when your input data is messy.

Single string resolution

The simplest case: one input, one canonical ID. Returns a confidence band and the surfaced matches so you can decide whether to accept automatically or queue for review.

curl -s "https://entidex.com/api/v1/entities/resolve?q=Anthropic" \
  -H "Authorization: Bearer $ENTIDEX_API_KEY"

Batch reconciliation

For lists of 25 or more, post the inputs as a single batch. Each input returns its own confidence band; apply a threshold (we recommend ≥ 0.75 for autopilot) and queue the rest for review.

curl -s -X POST "https://entidex.com/api/v1/entities/resolve/batch" \
  -H "Authorization: Bearer $ENTIDEX_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"inputs":["Anthropic","OpenAI","Mistral AI"]}'

Confidence bands

≥ 0.90 — exact match or canonical alias. Safe to accept automatically.
0.75 – 0.89 — strong match. Recommended autopilot threshold.
0.50 – 0.74 — likely match. Queue for human review.
< 0.50 — weak. Treat as unresolved.