AI for discovering evidence and knowledge

Generative AI is one layer. Six layers do the work.

Each layer has a different job. Chatbots appear in Layer 1 and language models in Layer 3, and neither works without the layers beneath them. Hover over or tap any example for a plain-language explanation.

LayerWhat it doesExamplesQuestion it answers
1

User Experience & Delivery

Where people choose what, where and how to search, and later receive the results.

Chatbots live here

WHAT? WHERE? HOW?People choose what to look for, which sources to search and which method to use.

2

Planning and coordination

Technical name: Agentic Orchestration

Plans the steps of a complex question and chooses which tools to use.

WHO? WHY?Interprets the person’s role and purpose, then plans the search.

3

Reading and understanding

Technical name: Model Intelligence

Programs that read, write, look at images and compare meaning.

LLMs live here

HOW?The matching and ranking behind the search choices in Layer 1.

4

Connecting knowledge

Technical name: Knowledge & Semantic Layer

Links information through relationships and agreed meanings, so it becomes knowledge.

WHAT?Defines the unit of knowledge, evidence, rather than raw data.

5

Storing information

Technical name: Data Foundation

One home for tables, documents, images and other kinds of information.

WHERE?Sources, their types and authority, indexes and database coverage.

6

Computing power

Technical name: Infrastructure

The hardware: computing, storage and networking.

All fourPowers every question above without answering one itself.

Read from the top: what people see sits above, and each layer depends on the ones below it. The question column maps each layer to the librarian’s search formula: Who? Why? What? Where? How?