Gleanor founder@gleanor.biz Index a first corpus

Docs

Datasets, profiles, and query routes.

Base URL https://api.gleanor.biz. Bearer tokens start with gln_live_. OpenAPI lives at /v1/openapi.json.

Quickstart

Create a project in the console, copy a test token, and hit the smallest route. Ujjwal Kumar Singh issues the first live token after a short call.

curl -s https://api.gleanor.biz/v1/health \
  -H "Authorization: Bearer gln_live_test"

Auth

Send Authorization: Bearer on every call. Rotate keys from the console. Test keys cannot touch production objects.

Indexes

Create a corpus, upsert documents, and publish a ranking profile. Idempotency keys on writes. List routes take cursor and limit. Indexing runs the sparse invert and, on Barn and above, the embedding pass.

POST /v1/indexes
POST /v1/indexes/{id}/documents
POST /v1/indexes/{id}/profiles
POST /v1/indexes/{id}/profiles/{pid}/publish

AI routes

Three query routes, three model jobs. Pass a published profile_id so Studio and production stay aligned.

  • Search — hybrid ranking. Keyword hits fused with embedding neighbors, then recency and field weights. POST /v1/search
  • RAG — cited windows sized for a generator you call. Filters on the request survive into the pack. POST /v1/rag
  • Recommend — item-to-item or user-to-item neighbors on the same corpus. POST /v1/recommend

Planned encoders and generators: Cohere Embed, GPT-4o, Llama 3 70B, AWS Nova. Access paths: Bedrock, OpenAI direct, Hugging Face. Gleanor does not run agents and does not execute the generator for you unless you ask for a hosted pack on the first corpus call.

curl -s https://api.gleanor.biz/v1/search \
  -H "Authorization: Bearer gln_live_test" \
  -H "Content-Type: application/json" \
  -d '{"index_id":"idx_demo","profile_id":"prf_live","q":"linen size guide"}'

Query

JSON in, JSON out. Errors use type, message, and request_id.

POST /v1/search
POST /v1/rag
POST /v1/recommend

Limits

Default 30 requests per second per key on paid plans. Burst tokens refill each minute. File a note to founder@gleanor.biz if a launch needs a higher ceiling.

Status

Incident notes go to founder@gleanor.biz. Webhooks retry for 24 hours with exponential backoff.