Search · RAG · recommendations

Glean the index. Rank with intent.

Gleanor is a search, RAG, and recommendations API. Engineers connect a catalog, help center, or wiki. Operators publish recency, tags, and synonyms from Studio. Production reads that same profile — embeddings, keywords, and weights on one corpus id.

One corpus search, RAG, recommend
Studio publishes what production pins
Hybrid keyword + embeddings + recency

Index a first corpus See where AI sits

query · catalog-spring

embed + keyword fuse

profile published

1
Spring catalog drop
Dense neighbor · merch weight 0.34
2
Warehouse FAQ
RAG cite · updated 4d ago
3
Size guide, linen
Field boost · title + tags

Embeddings related meaning, not just tokens

RAG pack cited chunks for a generator

Recommend neighbors on the same index

example index docs 2.4M p95 38ms

CatalogsHelp centersInternal wikisProduct feedsRunbooksMerchandising tagsSize guidesFAQs CatalogsHelp centersInternal wikisProduct feedsRunbooksMerchandising tagsSize guidesFAQs

AI in the product

Where models actually sit.

Gleanor is ranking infrastructure. Models embed documents, retrieve neighbors, and pack cited windows for a generator you choose. Operators still publish the profile. We are not shipping autonomous agents.

Use case · Hybrid ranking

Embeddings join the keyword score

For search engineers and merchandisers

Keyword hits and dense neighbors share a fused score. Cohere Embed is the planned dense encoder. Recency and field weights apply after the fuse, so Studio can bias a season without a new embedding job.

Workflow: connect source → embed + invert → Studio weights → POST /v1/search

Use case · Cited RAG

Retrieved chunks, sized for a generator

For support leads and help-center teams

Gleanor retrieves and packs cited windows. The generator is yours: GPT-4o, Llama 3 70B, or AWS Nova are the planned options. Filters on the ticket survive into the pack. We do not train that generator, and we do not run an agent over it.

Workflow: query + filters → retrieve → cited window → POST /v1/rag → your model

Use case · Recommendations

Neighbors from the same embedding space

For merchandisers and wiki owners

Item-to-item and user-to-item read the same corpus id as search. A catalog change is a related-item change. No second pipeline, no separate embedding store to drift.

Workflow: item or user id → neighbor search → POST /v1/recommend

Not an agent

Studio is a human publish step

For operators who own relevance

People draft a ranking profile, preview it on a saved query set, then publish. The API pin is that version. Gleanor is a retrieval tool — not an agent framework.

Workflow: draft → preview queries → publish → production pin

Planned model access: Bedrock, OpenAI direct, and Hugging Face. Exact encoder and generator are chosen on the first corpus call — we do not claim a live multi-model roster today.

Method

How Gleanor ships

Connect a source. Publish a ranking profile. Serve search, cited RAG, and recommendations from one corpus id.

Connect a source.

S3, webhooks, or a crawl. Document policies live on the dataset, not in a one-off script.

Publish a profile.

Operators set recency, tags, and synonyms. Engineers keep the query shape. Studio ships the version production pins.

Serve the index.

Search, RAG, and recommend share one corpus. A catalog change is a related-item change and a new citation source in the same hour.

Stack

What you actually buy

Ranking infrastructure with a studio merchandisers can use. Not a dashboard bolted onto a private index.

Keyword plus embeddings

Exact matches and embedding neighbors share a score. Recency and field weights apply after the fuse.

Cited RAG

Return excerpts with source ids, sized for a generator window. Filters on the query stay on the pack.

Recommend

Item-to-item and user-to-item from the same index. Related SKUs and related runbooks share a corpus id.

Studio

Operators draft a profile, preview it on a saved query set, then publish. The API pin is that version.

Desks

Who holds the knobs

Search engineers

Keep the query shape and the corpus id. Ranking weights live in the published profile, not in a fork of the service.

Merchandisers

Bias recency, tags, and synonyms from Studio. Seasonal boosts without waiting on a re-embed of the whole barn.

Support leads

Help-center RAG keeps the filter the ticket already set. Source ids travel with every excerpt.

Start

A working key in one afternoon.

Write founder@gleanor.biz. You get a test token, a sample project, and a 30-minute walkthrough on the catalog you actually want to rank.

Request a key