We have changed the company’s name. Straker Limited is now Arbitr Group Limited, effective 29th September 2026, and our shares now trade on the ASX as AGT. Nothing changes for you as a shareholder: same shares, same holding, same rights. You do not need to do anything. Our investor site is now investors.arbitr.ai.
Since listing, we have solved one problem: getting a company’s content into every language it operates in, accurately, at the scale a large enterprise needs. We did that for organisations where being wrong has consequences, in regulated industries and in regulated disclosure. That work earned us the customer relationships we have today.
AI has taken most of the difficulty out of producing content. A model will write anything you ask it to, in any language, in seconds. What it does not know is your company:
- which claims your legal team has already approved, or in which markets
- that a phrase you are allowed to use in one country is prohibited in another
- how you speak
- what you have already told the market and must stay consistent with
- what your strategy is this year
Every organisation has all of that. Almost none of it is written down anywhere a machine can use it. It sits in brand guidelines nobody opens, in a legal opinion in somebody’s inbox, and mostly in the heads of a handful of long-serving people.
That is why every piece of content still ends up in front of a person. It is not because people read better than machines. It is because people are the only place that knowledge lives.
That is the problem we build for now, and it is a different business from the one that solved the first problem. The company needed a name that said so.
Arbitr is our product, and now it is our company too. We store a company’s context: the rules it operates under, the claims and terminology it has approved, the way it communicates, its strategy, and the markets it sells in and what applies in each of them. Every decision one of its reviewers makes is kept, with their name and the date on it.
We then use that context to decide what can go out to market and what a person needs to look at first. The aim is that most content goes straight out. Anything that does not reaches a named person with the conflicting rule attached, so they are making a judgement rather than starting from scratch.
Context alone does nothing, though. It has to be worked: models that read it, and agents that apply it.
Having the product and the company share a name is deliberate. There is no longer a useful distinction between what we sell and what we are.
In case you missed Issue 7. On 1 July, our Japanese Subsidiary signed an AI partnership with IBM Japan, announced in Japan on 18 August. It is focused on exactly those two things: models and agents.
We contribute our own vertical language models, the reviewers behind them, and years of experience of how content goes wrong in regulated environments. IBM Japan contributes the infrastructure to run it and the governance layer around it. The platform uses watsonx Orchestrate to coordinate the agents and watsonx.governance to keep the record of what each one did.
This partnership matters more than a logo on a slide because large regulated customers will want a system like Arbitr to run on infrastructure and controls they already recognise. IBM is that standard. We have been an IBM build partner since 2023 (and are now a Platinum Partner). This takes the relationship from selling alongside each other to building together.
Thank you for your continued support.
David Sowerby
Interim Co-Chief Executive Officer, Arbitr Group Limited