Why we built Arbitr: a note from our Co-CEO
On September 29, Straker became Arbitr.
For more than 25 years, our customers have trusted us to help them solve a critical problem: how do we get our company’s content into every language we operate in accurately, and at the scale a large enterprise needs? We did this for organizations where getting it wrong matters, and that work earned us the customer relationships we have today.
That work still matters. But the question our customers bring us today has grown. It is no longer just “is this translation right?”. It’s “is everything we publish right, in every market, in every language, whoever or whatever wrote it?”
Our transformation to Arbitr is how we answer this new question, starting with the work we know best.
Getting it right is getting harder
Every business I speak to is feeling the same pressure. There is more regulation, more complex market dynamics, more channels and more languages. That means more ways to be wrong.
A price, a promotion, a product claim, a contract term or a disclosure has to hold everywhere it appears. Today, the things a company has approved live in different systems, with different people, and in different places. Keeping all of it accurate and consistent is slow, manual work that rarely keeps up.
At the same time, AI could take on far more of that work. Most leaders I meet want it to. What holds them back is simple: can they trust what it says? Wherever a business is with AI today, that trust decides how far it can go.
That’s why we built Arbitr. It governs what your AI knows, and checks what it says.
You approve it once. Arbitr applies it everywhere.
Today, that starts with translation.
Arbitr is an AI Context Platform. It translates your content, checks it against what you have already approved, and scores the result before it goes live.
At its heart is Cortex, your own memory of approved translations, terminology and the corrections your reviewers make. Your teams work in the web app, or send jobs from their own systems and AI assistants through our API, command-line tool and MCP server.
Every translation is checked against your glossary, memory and style rules, and each sentence is scored before it goes live. Low-scoring sentences are flagged, so your reviewers know exactly where to look first.
You decide when people step in. For high-stakes content, you can send the job to our linguists or your own reviewers, in the same platform. Their corrections go back into Cortex, so each job is more consistent than the last.
The result is that AI can take on more of the work, and you stay in control of what goes live.
Where we’re heading
Translation is where we start, not where we stop. Our goal is for Cortex to hold everything your business has approved, from product terms to claims and policies, so everything you publish can be checked against it, whoever or whatever wrote it. We will share each step as it is ready.
To the customers who brought us here
If you came to us for translation and localization, the expertise you rely on is still here. The same experts, the same care, now in one platform your teams can use directly.
What changes is what that expertise can do for you. You can import your translation memories and glossaries into Cortex, and every reviewer correction is added back. Each translation is checked against what you have already approved, not just run through a machine.
Our 37,000 expert linguists across 120+ languages stay at the center of this. The difference is that they start with the sentences most likely to be wrong, rather than reading everything.
Arbitr can be used by one team or many. For some of our customers, it will start with the localization team. For others it may be product, operations or head office marketing.
Why we can do this
Checking all your content in one language is hard. Checking it in every language is much harder. We know how far a translation can move a claim, and how to check it against what was approved.
We work closely with IBM as an IBM Platinum Partner, and our work already runs at enterprise scale: more than 10,000 IBM employees use our Slack agent, translating 40 to 50 million words a month.
The early response has been encouraging. As the VP of Product at Foxit Software put it:
“Arbitr gave our customers better outputs with greater precision. To say I am impressed is an understatement.”
Arbitr helps businesses publish with confidence in every language and move faster with AI, with people firmly in control of what goes live. We would love to show you what it can do. See Arbitr in action: book a demo.
David Sowerby, Co-CEO, Arbitr
