curiosity

Whitepaper · Search architecture

Modern enterprise search

The main kinds of search used in enterprise systems, how each works, and where each belongs. Written for the architects who have to choose.

For Architects and engineering leads Chapters 6, and a glossary Pages 21 Date October 2026

What it covers

Search is not one thing

Most arguments about search are about which layer is meant. A part number needs an exact match. A field report needs meaning. An agent needs both, and the permissions of the person asking.

  1. Layer 01

    Keyword

    Exact terms, BM25

  2. Layer 02

    Fuzzy

    Typos, variants

  3. Layer 03

    Vector

    Meaning, not wording

  4. Layer 04

    Hybrid

    Both, one ranking

  5. Layer 05

    Graph

    How records relate

  6. Layer 06

    Agents

    Search as a tool

Chapters

Six chapters, keyword to agents

01 · Page 03All six layers

Search is not one thing

Why most disagreements are about which layer is meant.

02 · Page 06Keyword

Where keyword still wins

Exact retrieval for part numbers and identifiers.

03 · Page 09Vector

What vectors add

Meaning beyond the words a field report uses.

04 · Page 11Hybrid · Graph

Hybrid and graph retrieval

Ranking that knows how records relate.

05 · Page 13Agents

Agents that search

Search as a tool, under the user’s permissions.

06 · Page 15All six layers

How Curiosity implements it

All six layers, in one system.

The full paper

The paper takes the layers apart, so you can choose them on purpose.

Download the PDF

Your data. Your infrastructure.

Connected knowledge for AI systems