Welcome to Hyeonggeun's Terminal!
2026. 10. 26. 4:52:36 AM

← Back to projects
$>

2025–present · #product · #beta

Companoid Labs

Likert AI — LLM-based HCI and UX Research Platform

An end-to-end survey research platform for study design, literature-grounded measurement items, survey export, and statistical analysis.

About this project

I lead the R&D of Likert AI at Companoid Labs. The platform helps HCI and UX researchers develop constructs and measurement items, taking a study from research design through survey export and statistical analysis. It has launched in beta.

Research workflow and implementation

I built the platform with React and TypeScript, FastAPI, and PostgreSQL, deployed on Railway. Google and Typeform OAuth support connected workflows, while PostHog provides product analytics.

The LLM and retrieval pipeline uses Gemini, Redis, and a Qdrant vector database of prior HCI literature and validated measurement items. An MCP server exposes study design, item generation, and analysis capabilities so AI agents can work with the platform directly.

UX Remix research

The earlier UX Remix work, developed with Jinkyu Jang as part of the Never Enough Project, explores how LLMs and prior literature can support measurement design. Researchers describe a project, select relevant constructs, and adapt measurement items to their evaluation context. The paper was presented at the Meta-HCI workshop at CHI 2025 in Yokohama.

The images above show the current Likert AI product and the earlier UX Remix research prototype from the 2025 paper.

Try Likert AI · UX Remix paper · Workshop

← Back to projects