---
schema: 1
type: term
title: '**bottom-up**'
description: >-
  **bottom-up** — an approach to exploring a subject that starts from concrete
  details and sources and builds broader structure from them, rather than
  beginning with a pre-existing framework and fitting the details into it.
canonical: 'https://liquidgold.cc/channel/everyinc/explore/term/bottom-up-a2994e6713a6'
markdown: 'https://liquidgold.cc/channel/everyinc/explore/term/bottom-up-a2994e6713a6.md'
llms: null
parent: 'https://liquidgold.cc/channel/everyinc/llms.txt'
channel: everyinc
updated: '2026-10-23T22:12:35.346Z'
language: null
children: []
---


# \*\*bottom-up\*\*

Canonical: https://liquidgold.cc/channel/everyinc/explore/term/bottom-up-a2994e6713a6

Parent: [Channel index](https://liquidgold.cc/channel/everyinc/llms.txt)

## Definition

**bottom-up** — an approach to exploring a subject that starts from concrete details and sources and builds broader structure from them, rather than beginning with a pre-existing framework and fitting the details into it.

## Where the term shows up In the material on **Learning, Reading, and Research with AI**, the term *bottom-up* marks a direction of travel for exploration: start from the concrete details in front of you — your sources, notes, and documents — and let broader structure emerge as you go, rather than beginning with a finished framework and fitting details into it. This sits alongside the channel's core workflow advice, which is to **ask AI for structure first, then do the understanding yourself**. The two ideas are complements, not opposites: AI can give you a provisional skeleton, but the bottom-up work of reading and connecting the actual material is what makes the structure yours. ## How to explore the term When you explore a new subject with AI as an active partner, you can treat bottom-up as a discipline check on any top-down scaffold: - **Does the outline the AI offered match your sources?** If not, adjust the outline to the evidence, not the evidence to the outline. - **Let details reshape the big picture.** A grounded notebook — the kind discussed in the episode with Steven Johnson — makes this easier, because the assistant stays anchored in your own documents while you build understanding from them. - **Recognize the limits.** A chat model can propose structure quickly, but only the bottom-up pass through the material tells you whether that structure holds.  - **Bottom-up** = building understanding from concrete sources and details upward, instead of starting from a pre-imposed framework. - It complements, rather than replaces, asking AI for structure first. - Grounded research tools keep the bottom-up pass anchored in your own documents.