QACAT

Catch translation issues before your users do

Languages

Developer Tools

Artificial Intelligence

Visit WebsiteSee on Product HuntTwitter ⧉

Scoring per ai engine is useful. Most teams try multiple ai engines, and knowing which engine performs best per language is valuable data.

Well done Dmitry - the feature choices show this platform was built by someone with a lot of hands-on localization experience =) agree that handling glossary terms is a must! when I was working on Nitro, glossary feature was often asked about, so we added it too.

Hi! I am wondering how this is different from just running a quick LLM QA check myself?

Congrats on the launch! You've put a lot of work into this platform. I personally like that you can pick how deep the QA goes - makes sense that not everything needs the full treatment

Awesome.. where have you been. We are localized in I think around 9 countries and theres no room for error because our product is used for a lot of legal tech. Passing to my team.

How does the OCR handle languages with non-Latin scripts or right-to-left text like Arabic or Hebrew?

finally a QA tool that treats quality as something you track over time rather than a one-off score per project!

Congrats on launching QACAT, Dmitry! Love that there is an nda safe automated layer with no AI involved - a lot of teams can't send content to LLMs so having a deterministic-only option is useful indeed

The screenshot review with OCR caught my eye. I've seen how time linguists waste when doing LTQ and re-typing strings and indicating what’s wrong, so this feature sounds really handy.

How does it perform at scale? Wondering if it's been tested on projects with, say, 40+ languages.

Hey Product Hunt.

Dmitry here. I lead linguist operations at Alconost — my job is making sure our linguists' work is as efficient as it can be.

For a long time, the translation QA process seemed okay. Spreadsheets, marked errors, a final score. Workable for most projects, awkward for screenshot testing — but there was no real alternative.

We've come to the point where QA is often ditched by teams just because it isn't adapted to modern translation workflows. The good old QA in spreadsheets is slow, inconvenient, and more expensive than it could be.

AI made translation cheap. It also made quality invisible.

QACAT is what we built to make quality measurable again — and to give linguists a real environment to do real work in.

A few things make it different:

– Pick the right QA depth for the content. Rule-based checks (fast, NDA-safe); QA by pure AI or AI + human review; full human evaluation — all on one platform.

– Reports, not spreadsheets. Every QA run produces a structured report — score, severities, error categories, language and engine breakdowns. When it's done, an automatic summary highlights risk areas and points to what actually needs fixing.

– A real environment for the people doing the work. Upload screenshots, and the platform handles the rest — OCR reads the text, the translation auto-fills the correction field, and reviewers mark issues right on the UI. No external image editors, no link-pasting, no re-typing strings.

– Quality you can track over time. Linguist performance, severity patterns, recurring issues — across projects, languages, and people.

Would genuinely love your feedback — what's missing, what's confusing, what you'd want to see next. We'll be here all day answering questions.