RAG stack: built vs buy — Glean + Notion Enterprise Search vs Cohere + Nomic open-weight

5 sources

Enterprise search (Glean 5 + Notion Enterprise Search) and retrieval models (Cohere + Nomic) sit at different layers of the RAG stack — enterprise search is the all-in-one surface that ships connectors + permission + agents; retrieval models are the open-weight / API building blocks that go into a custom RAG pipeline. The two are not equivalent: enterprise search is buy (managed surface), retrieval models are build (custom pipeline). Creators choosing between them pick based on whether they want a managed surface (Glean / Notion) or a custom pipeline (Cohere / Nomic).

Why now

All four shipped in the same week — the RAG stack just got a clear buy / build split between enterprise search surfaces and retrieval model surfaces.

Why it is worth publishing

Demo potential: side-by-side of a RAG workflow built with Glean + Notion Enterprise Search (managed) vs Cohere + Nomic (custom).

Evidence basis

Four independent vendor primary sources + The Decoder weekly roundup

“Four RAG vendors shipped updates in one week — Glean + Notion (managed enterprise search) vs Cohere + Nomic (retrieval models for custom RAG) — and the RAG stack just got a clear buy vs build split.”

Angle

Frame the four as 'buy (managed enterprise search) vs build (custom RAG pipeline)' — pick based on whether the creator wants a managed surface or a custom pipeline.

Format

Long-form explainer

Demo idea

Record a 10-minute explainer: 3 min intro on 'buy vs build' framing, 3 min on managed (Glean + Notion), 3 min on build (Cohere + Nomic), 1 min on the decision matrix.

Platform notes

Per-tier pricing and connector coverage beyond the captured summary are not extracted; confirm specific surface details against the underlying vendor docs before stating them on the record.

Usable claims

Glean shipped 5 — enterprise search with permission-aware connectors (respects source-system ACLs), Glean Agents for workplace automation, Glean Assistant for natural-language data interaction.
Notion shipped Enterprise Search GA (search across Notion + Slack + Google Drive + GitHub + Jira + Linear + Asana + Salesforce + Zendesk + Box) and AI-for-every-database.
Cohere shipped Rerank 3.5 (improved retrieval accuracy, multilingual support, multi-vector support) and Embed 4 (1024-dim default embeddings with multilingual and image-text alignment).
Nomic shipped Atlas Embed 2 — long-context embeddings (8K token context per embedding call), open-weight license, Atlas visualization dashboard.

Evidence pipeline

From the news

Breakdown

Enterprise search (Glean + Notion) sits at the managed-surface layer; retrieval models (Cohere + Nomic) sit at the custom-pipeline layer. This explainer keeps the two layers distinct, rather than treating 'RAG stack' as a single category.

Sources

Risks

Use The Decoder and IT之家 as media-type corroboration, but read the underlying vendor docs for any specific accuracy claim or connector list before stating it on the record. Verify specific capability claim against the underlying vendor docs and the actual license / pricing matrix before stating it on the record; do not paraphrase per-platform pricing or license terms into specific dollar figures or commercial-use clauses.
Vendor docs confirm feature existence but specific pricing beyond the captured summary is not in this pass. Verify specific capability claim against the underlying vendor docs and the actual license / pricing matrix before stating it on the record; do not paraphrase per-platform pricing or license terms into specific dollar figures or commercial-use clauses.
Vendor docs confirm improved accuracy claims but specific benchmark numbers are not in this pass. Verify specific capability claim against the underlying vendor docs and the actual license / pricing matrix before stating it on the record; do not paraphrase per-platform pricing or license terms into specific dollar figures or commercial-use clauses.

Demo ideas

Side-by-side RAG workflow: same workflow built with Glean + Notion Enterprise Search (managed) vs Cohere + Nomic (custom), measure time-to-deploy and control surface.
Decision matrix: managed vs custom — plot each surface on time-to-deploy vs control vs vendor lock-in.
Migration story: 'migrate from custom RAG to managed enterprise search', measure porting work and surface delta.