About

Reading Companion is an AI-scaffolded reading tool — and, behind the scenes, a research instrument.

The idea

Most reading tools either get out of your way or take over. We wanted a third option: a companion that helps you stay in the text. The AI here is built to ask questions rather than write your answers. It probes for evidence, mechanism, and interpretation; it refuses to summarize; it acknowledges what you noticed and asks what you might notice next. It is meant to deepen attention, not to replace it.

Who made it

Reading Companion is a collaboration between Jewoong Moon (University of Alabama) and Joomi Lee (University of Arkansas). It is being developed for a two-study program of research on AI-mediated reading, grounded in the Limited Capacity Model of Motivated Mediated Reading (LC4MP), Interactive-Constructive-Active-Passive engagement (ICAP), and a cross-cutting construct we call Locus of Regulation in AI-Mediated Reading.

What it measures

The tool emits a timestamped telemetry event for every meaningful action — section enter, scroll, idle, paragraph dwell, highlight, annotation, checkpoint response, AI intervention — so reading-time behavior can be analyzed alongside the reflection notes that come out of each session. No personal identifiers are required to use the tool.

The two studies and their analytical layers are described in the team's public Research Plan. If you're a researcher interested in the data or methods, get in touch.

Open source

The codebase is on GitHub: github.com/Educatian/reading-companion. Issues and contributions welcome.