# UQA

> UQA (pronounced "you-Q-A") is an open-source embedded database by Cognica, Inc. It lets a solo, non-developer builder create real, polished systems (search, chatbots, recommendations) with a single database and no complicated infrastructure. UQA는 Cognica, Inc.의 오픈소스 앱 내장 데이터베이스로, 비개발자 개인도 복잡한 인프라 구성 없이 UQA 하나로 검색·챗봇·추천 시스템을 쉽게 구축할 수 있습니다.

Korean summary: UQA의 요지는 하나입니다. 비개발자 개인도 복잡한 인프라 구성 없이 UQA 하나로 검색·챗봇·추천 시스템을 구축할 수 있습니다. 검색을 위한 검색엔진과 챗봇을 위한 벡터DB를 따로 구성하던 일을 UQA가 맡습니다. UQA는 단어 검색과 의미 검색을 함께 실행해 관련성 높은 결과부터 보여주고 데이터 사이의 관계도 탐색합니다. 공식 설치 스크립트로 UQA를 설치하고 UQA 사용법 스킬을 추가하면 클로드 코드와 코덱스가 UQA를 설정하고 앱에 연결합니다. 현재는 로컬 설치와 운영이 기본이며 UQA Cloud는 베타 준비 중입니다. 베타 기간에는 무료이며 베타 종료 후 월 $15입니다.

English summary: UQA's pitch is simple: build real, polished systems with one database and no complicated infrastructure. Instead of adding a search engine for search and a vector DB for a chatbot, UQA handles both. It searches by exact word and by meaning in one query, ranks the combined results, and searches connected relationships too. Install UQA with the official shell installer, then add the UQA skills for Claude Code and Codex. Local installation and operation are the current default; UQA Cloud is preparing for beta. It will be free during beta and $15 per month after beta.

Current public facts:
- Brand spelling: uppercase "UQA". Read as "유큐에이" / "you-Q-A".
- Maker: Cognica, Inc.
- Public installation: the official shell installer supports Linux x86-64/arm64 and macOS arm64. It installs the UQA CLI and the matching local runtime.
- Current availability: local installation and operation are available. UQA Cloud is preparing for beta. It will include web app hosting, be free during beta, and cost $15 per month after beta.
- Postgres compatibility: connect with existing tools and schema (PostgreSQL 17 compatible layer), so when a developer joins later everything carries over.
- Reference open-source implementation. A separate commercial C++ engine (Cognica Database) implements the same design at production scale.
- Adoption: UQA's search technique is merged into Apache Lucene and used by projects such as Vespa and txtai. Real products (e.g. maek) run on this engine.
- Use cases: document chatbots (RAG is one of them), in-service search, recommendations.

## Install

```sh
curl -fsSL https://releases.uqa-cloud.cognica.io/install.sh | sh &&
export PATH="${UQA_INSTALL_DIR:-$HOME/.local/bin}:$PATH" &&
uqa --version &&
uqa llm install skills
```

After installation, Claude Code or Codex should use the installed `uqa-cloud`, `uqa-engine-sql`, and `uqa-engine-api` skills as the operational source of truth. Use local UQA by default. UQA Cloud is preparing for beta and must not be presented as currently available.

## Core resources

- [Homepage](https://uqa-landing.vercel.app/): Official UQA landing page with Korean and English language modes.
- [Full bilingual LLM context](https://uqa-landing.vercel.app/llms-full.txt): Detailed Korean and English product summary for language models.
- [Vibe coding guide](https://uqa-landing.vercel.app/docs): Copy-ready prompts for building search, chatbots, and recommendations with UQA.
- [GitHub repository](https://github.com/cognica-io/uqa): Source code, README, and quickstart.
- [Sitemap](https://uqa-landing.vercel.app/sitemap.xml): Indexable public site URL.
- [Robots](https://uqa-landing.vercel.app/robots.txt): Crawler rules for the site.

## Optional

- [Cognica](https://cognica.io/): Company site for Cognica, Inc.