Tools & Integrations

Data Streams & Export API

Connect Reaudit to your data warehouse and CDP — push AI visibility events to RudderStack, Segment, or webhooks, and pull warehouse-ready tables via the Export API.

Data Streams connects Reaudit’s AI visibility data to your data stack. There are two complementary modes: event destinations push events (visibility score changes, brand mentions, citations) to RudderStack, Segment, or any webhook in real time, and the Export API lets your warehouse pull flat, cursor-paginated tables (answers, citations, visibility stats, attributed sales) on your own schedule. Both are configured from Tools → Data & Warehouse.

Event Destinations (push)

  1. 1Open Tools → Data Streams and click "Add Destination"
  2. 2Choose RudderStack (data plane URL + source write key), Segment (write key), or Generic Webhook (URL + signing secret)
  3. 3Select the events to stream — aggregate events like "AI Visibility Score Changed" are on by default; answer-level "Brand Mention Detected" is opt-in because it can be high-volume
  4. 4Click "Test" to send a synthetic event and verify the connection
  5. 5Watch the per-destination delivery log (status, latency, retries) to monitor health
  • Format — Events use the Segment track spec, so RudderStack/Segment route them to any downstream destination or warehouse without transformations
  • Webhook security — Payloads are HMAC-SHA256 signed via the X-Reaudit-Signature: t=,v1= header; verify with hmac_sha256(".", secret)
  • Reliability — Deliveries retry automatically (3 attempts with backoff) and never slow down the Reaudit app; failures appear in the delivery log
  • Identity — The Segment userId is a stable hashed account identifier, not personal data

Export API (pull)

Create a bearer key in Tools → Data Streams → Export API (keys start with rdk_ and are shown once). Then pull tables from /api/export/v1/ in JSON, NDJSON, or CSV. GET /api/export/v1/schema returns the full table catalog so Airbyte, Fivetran custom connectors, dlt, and reverse-ETL tools can introspect it.
  • Tablesai_answers, citations, prompt_visibility_stats, visibility_gaps, visibility_lift, attributed_sales, llm_usage
  • Pagination — Opaque cursor over (timestamp, id); follow next_cursor (JSON) or the X-Next-Cursor header (NDJSON/CSV)
  • FiltersprojectId, since, until, limit (max 1000 rows per request)
  • Rate limit — 60 requests/minute per key
  • Privacy — End-user PII (customer emails, names) is never exported; tables are brand/answer/aggregate level
bash
curl -H "Authorization: Bearer rdk_..." \
  "https://reaudit.io/api/export/v1/ai_answers?since=2026-01-01&format=ndjson"

Warehouse loading pattern

Schedule a nightly job that calls the Export API with since = your last watermark and format=ndjson, streams rows into a staging table, and merges on id. The stable cursor guarantees no gaps or duplicates.
data streamsexport apiwarehouserudderstacksegmentwebhooksnowflakebigqueryredshiftairbytereverse etlcdpetl