Press kit
Taia: so a company sounds like itself in every language
AI translation, collaboration tools and professional linguists, with one memory and one glossary across everything a company translates. 204 languages, 65+ file formats, and the formatting survives.
A company says the same things in a hundred places: the manual, the catalog, the app, the training video. In one language that's manageable. Add more, and the product name comes back three different ways and the tables fall apart.
Taia started in 2018 with documents, where that hurts most: whole files translated and returned looking like the original, in the company's own terminology. More than 30,000 people have signed up so far. Next come the words that live outside documents, the text in a company's app and the knowledge in its training videos, all sharing the same memory, so a term gets fixed once and stays right everywhere.
Latest
-
Press kit published: boilerplate, fact sheet, logos, screenshots and a press contact in one place.
-
Waitlists open for Strings (app text) and for audio and video translation.
-
Case study: Quantum Connections scaled from 1 to 6 languages with AI and expert review.
-
Case study: author Carole Sainte-Marie translated a 63,000-word novel into English for $99.
What's coming
Taia started with documents. Two new things are on the way. Both will open as early access by invitation, through the waitlists below. Nothing here is public yet, so please describe them as coming, not live.
Strings: your app's button labels, in every language
Every release adds a few keys and changes a few more, and somebody exports them, translates them and pastes them back. Strings is Taia's answer: new and changed text will go to Taia, translations will come back, and only approved text will reach your build. We'll run it on our own app and help center before anyone else gets it.
Join the Strings waitlistAudio and video: the training video that stays in one language
Webinars, trainings and interviews hold a lot of what a company knows, and they're the content most likely to stay in one language. We're building transcription and translation for recordings, so a video becomes one more file in your Taia account.
Join the audio and video waitlistOne memory for all of it
Documents, app text and recordings will share one translation memory and one glossary. Get a term right once, and it stays right in the manual, in the app and in the video.
How translation memory worksStory ideas for journalists
Seven angles we can back up with real products, real customers and a demo account. Ask for any of them, or tell us the one we missed.
-
Angle 1
How much of AI translation do linguists actually change?
Across about 211,000 segments (2 million words) that professional linguists post-edited on Taia in 2026, 54% of segments needed no change and linguists changed about 15% of the characters. Weighted by words, 37% of the text was untouched. Standard AI engine, 30 customer accounts. Short segments are the ones most often left alone, which is why the word figure sits so far below the segment one. We can share the full method on request.
-
Angle 2
One feature, three names
The same feature gets one name in the manual, another in the app and a third in the training video, and nobody notices until a customer does. Terminology is the quiet cost of translation. A glossary and a translation memory fix a term once for documents, and Taia is extending the same memory to app text (Strings) and to audio and video, so it will stay fixed everywhere. The video below shows it happening in documents, at Volta Air, a flying car company we made up for our demos.
-
Angle 3
What a small company learns translating its own product with its own tool
Before anyone else gets Strings, we'll run it on our own app and help center. What breaks, what turns out to be easier than expected, and what it's like to be the customer for once.
-
Angle 4
AI plus humans, not AI instead of humans
Novelist Carole Sainte-Marie translated a 63,000-word novel with Taia and kept 94% of the AI output unchanged, after reading all of it. Taia doesn't pretend the AI is finished: professional linguists work inside the same project, in an ISO 17100 certified process.
-
Angle 5
Why AI translation still breaks documents
Translation engines got very good at sentences. Documents aren't sentences: they're tables, tags, headers, text boxes and fonts. We can run the same file through a plain text tool and through Taia, and show why layout is the hard part. Taia reads 65+ file formats and keeps the formatting.
-
Angle 6
The software company that nearly dropped languages
LightBurn Software was considering dropping to 6 to 8 supported languages because keeping up with translation was too much work. Today its software is localized into 27.
-
Angle 7
The manual that has to be right in every market
Companies that build and service machines and vehicles live on manuals, datasheets and training material that have to read correctly everywhere. Unior Special Machines cut its translation costs by 65 to 70%. Hyundai Avto Trade cut translation time for its mechanics' training materials by more than 60%. Both stories are published, with the customers' approval.
Want to try it before you write? Ask for a demo account, or a conversation with Matija, at press@taia.io.
About Taia
Use either version as written. Copy it, or quote from it.
Short, about 60 words
Taia is an AI translation platform that started with the documents companies actually have. Instant AI translation, collaboration tools and professional linguists live in one place. Founded in 2018 and headquartered in London, Taia is ISO 17100 certified and never uses customer content to train AI models. Next, it is extending its translation memory and glossary to app text and to audio and video.
Long, about 150 words
Taia is an AI translation platform for the documents companies actually have: manuals, catalogs, contracts. Upload a file in one of 65+ formats, choose from 204 languages, and get it back with the formatting intact and your own terminology applied. Instant AI translation, built-in collaboration tools and on-demand professional linguists live in one place, so a project can start with AI and move up to a human expert without changing tools. Translation memory and glossaries improve with every project. Pricing is transparent, with instant quotes and no hidden fees. Customer content is never used to train AI models. Founded in 2018 and headquartered in London, Taia has more than 30,000 registered users and more than 2,000 vetted native linguists, and is ISO 17100 certified. Next, Taia is extending the same translation memory and glossary to app text and to audio and video. Customers include Mozilla, Red Bull, Hyundai, 3M and Atlas Copco.
Fact sheet
- Company
- Taia
- Legal entity
- TAIA TRANSLATIONS LTD (UK company no. 12779985), incorporated 30 July 2020
- Founded
- 2018
- Headquarters
- London, United Kingdom
- Registered office
- 71-75 Shelton Street, Covent Garden, London WC2H 9JQ, United Kingdom
- Funding
- Angel investors at the start, venture-backed in 2020 and 2021, bootstrapped since
- Languages
- 204
- File formats
- 65+, with formatting and layout preserved
- Registered users
- 30,000+
- Vetted native linguists
- 2,000+
- Certification
- ISO 17100
- Data policy
- Customer content is never used to train AI models
- Products
- AI Translator, Translation Studio and Translation Editor, translation memory, glossary, professional linguist services, API
- Reviews
- 4.8 on G2
- Customers
- Mozilla, Red Bull, Hyundai, 3M, Atlas Copco
Logos
SVG for print and web, PNG for slides. Please keep clear space around the logo, and don't recolor, stretch or add effects.
Product screenshots
Click an image to open the full-size file. The product screenshots show a fictional demo company.
The people
Taia has been building since 2018. The in-house team is a small group of product people, project managers and in-house linguists, working from London and Ljubljana. More than 2,000 vetted freelance linguists around the world do the professional work, and when a project comes in, a project manager picks the linguists and stays with it until delivery.
Quotes you can use
Pick the one that fits your story. All are from Matija Kovac, co-founder and CEO of Taia.
On why Taia exists
"The internet promised that anyone could sell anywhere. Language is the part of that promise that still isn't kept. A ten-person company should be able to sound as clear and trustworthy in Japanese as it does at home, without hiring a department to do it. That's the company we're building."
On AI and trust
"Turns out translation was never really about words. It's about whether someone in another country trusts you enough to buy from you. AI made the words cheap. What we're building is the trust: the right terms, the layout intact, and a human expert when it counts, for everything a company says."
On where this is going
"In ten years, nobody will ask whether a company works in other languages, the same way nobody asks whether it has a website. It'll just be normal. We're building the part that makes it normal: AI that keeps a company's own voice, and people who check it when it matters."
On one voice
"Most companies already speak a dozen languages, just not with one voice. The manual says one thing, the app another, the video a third. We think a company should sound like itself everywhere, in every market, and that should be something a small team can do on a Tuesday afternoon."
Matija Kovac (Kovač), Co-founder and CEO
Matija co-founded Taia in 2018 and is its CEO. He runs sales, product and strategy, and still ships code himself.
His focus is how AI translation, professional linguists and translation technology fit together, and how companies that sell across borders can translate their manuals, catalogs and contracts without losing their terminology or their formatting. He writes about machine translation and the localization industry, and makes videos about how translation really works.
Customer stories
What changed for the people who use it. Every story is published on taia.io with the customer's approval.
- Carole Sainte-Marie, novelist Fiction, French to English A 63,000-word French novel into English in one month, with 94% of the AI output kept as it was. Read the story
- Best Secret Fashion retailer Translation orders that took up to a week now come back in 2 to 4 days. Read the story
- Hyundai Avto Trade Automotive Translation time for mechanics' training materials down by more than 60%. Read the story
- Medis Pharmaceutical About 30% less translation time, and more consistent terminology, within a 14-day trial. Read the story
- LightBurn Software Software Weighed dropping to 6 to 8 languages, and is now localized into 27. Read the story
- Quantum Connections Dialogue training From one translated course to six target languages. Read the story
- Unior Special Machines Machine manufacturing Translation costs down 65 to 70% against the previous language provider. Read the story
- Millpledge Veterinary Veterinary products 80% less admin time, and a quote within minutes of uploading. Read the story
Watch
Short, public and a little silly on purpose. The company in the demos is Volta Air, a flying car maker we made up. Videos load only when you press play.
AI document translation that keeps your formatting (5:07)
The five-minute overview: one file, many languages, and the layout survives.
Why your translations call the same thing by 3 different names (3:38)
Glossary, translation memory and style guide, and how a word gets fixed once.
This flying car manual needed 11 languages by Friday (0:35)
Copy, paste, paste back, fix the table. Now do it 10 more times (0:28)
How long does it take to translate a manual into 11 languages? We timed it (0:25)
Press contact
For interviews, quotes, review access or anything this page doesn't cover, write to press@taia.io with "Press" in the subject line. We reply within one business day.
Registered office: 71-75 Shelton Street, Covent Garden, London WC2H 9JQ, United Kingdom