Martin Calvert on AI, Localisation and the Future of International iGaming Growth

After seven years at ICS-digital, what made this the right moment to join All-In Global, and what specifically drew you to CRAIG?

Seven years at ICS - and nearly three years at Blueclaw before that - gave me a lot of knowledge about where iGaming brands can experience ROI from translation and multilingual content, but it is the 2026 context that made this move timely.

Challenges around speed, cost and quality aren’t new, but these days the level of competition brands are facing is intense and this is coupled with tax pressures, regulatory scrutiny and more mature markets defined by players with higher expectations.

I think in this context, the CRAIG AI translation platform combined with All-In’s subject-matter-expert human QA and optimisation is a big part of the answer to these challenges - a kind of ‘best of both worlds’ approach that allows brands to move faster, with quality and at a very very very attractive price.

I also like that CRAIG, while a mature product, is still developing and evolving. I can spend time with the innovation team, clients and partners, find out what people actually need and feed that back into how we develop and position the proposition.

Where do you believe AI genuinely improve localisation, and where does human expertise remain essential, especially given how small mistakes can carry commercial or regulatory consequences?

If a product team has 30,000 strings, a studio has an enormous catalogue or a marketing team is maintaining content across ten markets, there is a huge amount of work where AI can be genuinely useful, particularly for speed, consistency and the ability to process volume - if the right safeguards are in place and the platform has industry-specific optimisation.

I’m not convinced having a specialist linguist repeatedly translating near-identical strings or hundreds of subtly different pieces of ad copy is necessarily the highest-value application of their expertise, and it also basically invites mistakes due to fatigue.

The best approach - as far as I can see - has specialist linguists helping shape terminology, brand rules and the workflow of translation projects, with human judgement concentrated on the places where it actually adds something like the edits, quality assurance and polish.

CRAIG operates alongside an extended team of more than 500 native linguists and offers human post-editing and QA around the technology, but the input of client-side teams can be hugely valuable too.

What problem is CRAIG solving that general tools like ChatGPT or standard machine translation don't solve well for iGaming specifically?

The first thing is to acknowledge that ChatGPT, Claude, Gemini and other general platforms can translate things very well - especially when it comes to simple meaning or understanding.

Lots of very smart marketing, product and localisation people are already getting useful work out of these tools, but the question changes when you go from “translate this paragraph” to running localisation continuously across products, teams and markets that is not just accurate, but attractive, while also being on the right side of regulations and brand guidelines.

To manage translation cost-effectively, brands need terminology to persist, brand language to remain consistent, content to fit game/platform interfaces, files to move through sensible workflows and different material to receive different levels of human scrutiny, and this is all built into the CRAIG workflow.

I think you can absolutely build a sophisticated system around a general AI model to do much of that, but to me that sounds extremely difficult if you don’t have dedicated engineering resource, localisation expertise and team continuity/appetite to build and maintain it.

CRAIG's appeal is that this is the problem it was specifically designed around. It supports more than 50 languages, client-specific terminology and brand voice, API workflows and dedicated linguistic and engineering support.

So - in short I don't think the interesting comparison is “ChatGPT can't translate and CRAIG can”. It is whether you want a very capable general tool and to construct the localisation operation around it yourself, or something designed from the outset to behave like part of that operation, with the human linguists seamlessly integrated into the workflow for QA.

What separates content that's merely translated from content that can actually compete in a local search market?

Quite a lot, unfortunately for anybody hoping international SEO can be reduced to pressing a ‘deliver global traffic’ button.

The most basic problem is that people in different countries rarely search for the same things in the same way using equivalent words and, at the same time, many casino/table games have the same types of mechanics and features, while on the sportsbook side everyone has more or less the same odds and, of course, the same matches/games to work with.

Marketers need to look at the actual search engine results pages, AI Overviews, competitors, terminology, product preferences, regulation, publishers and what somebody in that market is really trying to accomplish when they search - bearing in mind that what might be a ‘good’ answer last month may not be the best ‘this’ month.

Sometimes translated copy is absolutely fine, but sometimes the whole emphasis of the page needs to change. It all comes down to what is going to be seen by users (and search engines) as Helpful Content - a term Google loves - and if that matches customer intent at the right time, backed up by the authority of a strong brand, link building, earned media coverage and so on.

International SEO starts with researching the local opportunity, rather than translating the domestic strategy after it has already been decided.

Brazil is a good contemporary example. The SERPs are intensely competitive, national media have substantial visibility and there are interesting gaps around creators, AI visibility and genuinely useful sports content.

You have to look at what is happening in that ecosystem and who the current winners and losers are to evaluate if you can compete at all, rather than arriving with an English-language plan and a Portuguese dictionary.

A page can be linguistically immaculate and rank absolutely nowhere if it doesn’t align to local search intent, or the surrounding site is weak.

This is where I see CRAIG having special significance where, if brands can refresh the English version of their content for specific markets, they have the reassurance of knowing what they’re saying in a certain market and retain control of their messaging and offers.

At the same time, they have the security of knowing that the translation part can be done quickly, accurately and then optimised further by human linguists.

With Google search changing rapidly and AI-generated answers taking up more space, what does a strong international SEO strategy look like for 2026 – 2027?

Traditional search continues to drive an enormous amount of commercially useful traffic, but of course that doesn't mean we ignore what is changing around it.

Google increasingly puts assembled answers in front of conventional results and marketers also have to think about LLMs, Discover, social platforms, video, publishers, creators, forums and communities.

For me that makes international SEO broader, but it doesn't make the fundamentals disappear.

You still need sound technical SEO, good architecture, proper local keyword and competitor research, genuinely useful material and some reason for people to trust you.

That overlaps with the way I've thought about digital PR since we just called it ‘content marketing’ back in 2015. Outreach and media coverage should at least aim to do more than earn a link.

Topical relevance, the readership, whether the coverage itself ranks and whether anybody relevant actually encounters the brand are all considerations that can result in campaigns delivering more value.

AI makes producing thousands of pages cheap. Does that risk flooding iGaming with interchangeable content, and how can brands avoid that?

Absolutely, but I think a lot of companies have been guilty of that for years, where they’ve churned out undifferentiated human content.

These days with AI, brands are even more well equipped to produce enormous quantities of content nobody particularly asked for.

iGaming is especially vulnerable because there are so many repeatable formats: casino reviews, bonus pages, payment-method pages, game descriptions, operator comparisons and endless permutations of “best betting sites”.

A model can produce 10,000 pages without hassle. Unfortunately it doesn't produce 10,000 new reasons for anybody to read them.

If ten competitors use broadly similar models, source material and prompts to answer the same query, there is a fairly obvious risk that everyone ends up with superficially different versions of much the same thing.

I wouldn't respond by becoming puritanical about AI for this type of informational content, but the question I would keep asking is what the content contains that actually deserves to exist. First-party data? Real product testing? Local expertise? A useful tool? Original research? Somebody with actual experience prepared to give an opinion?

AI makes generic content much faster to produce, but it doesn’t help you displace a competitor with comparable content who is already in the top spot. The goal should be how to improve upon what is already there, or provide an alternative (but valuable) point of view that audiences will value.

I think thoughtful AI use can help close the gap there and, with regard to CRAIG, scaling internationally becomes a lot more cost-effective and achievable for brands who may have hit a ceiling in their SEO progress in their home countries.

How should AI systems handle jurisdiction-by-jurisdiction differences in bonus claims, responsible-gambling wording, or financial terms without creating extra regulatory risk?

Very, very carefully, is the unglamorous answer to this question, but more than this, human oversight is needed and human decision-making about appetite for risk.

Compliance and legal teams establish what the requirements are, what claims are permissible and which wording has been approved. Technology can then become extremely useful in helping apply those decisions consistently across a lot of material.

Where clients have approved terminology, brand language or market-specific wording, those requirements can form part of the localisation workflow rather than somebody repeatedly attempting to explain them in a fresh prompt.

CRAIG is designed around persistent client terminology and expert-guided customisation, and with human linguists and strategists we can train, repeat and check outputs so that they match client guidelines.

Which iGaming markets or languages are currently underserved from a localisation standpoint?

I think the gap becomes particularly interesting in newly regulating markets and in markets that are commercially attractive but not yet large enough for every operator to maintain an extensive native-language team internally.

In these cases there are considerably fewer people who combine language skills and translation credibility with iGaming, regulation, product and digital marketing knowledge. That's where “underserved” becomes more meaningful to me and, of course, is an area we actively tackle with how CRAIG has been optimised for even ‘uncommon’ languages.

Looking three years out, do traditional translation workflows disappear, or does it become a hybrid model where AI handles scale and specialists become editors and cultural advisers?

I expect linguists to spend more time as editors, subject-matter specialists, cultural advisers and quality owners, and less time being valued primarily according to how many words they can manually move from language A to language B.

CRAIG already reflects some of that hybrid direction, combining AI with native linguists, post-editing and QA.

Genuine subject expertise, judgement and cultural knowledge become even more valuable as the ‘easy’ parts become easier to automate.

If a CEO said “we want to enter five new markets using AI to control costs,” what's the first mistake you'd warn them against?

I'd want to know why those markets were selected in the first place. What's the actual opportunity? How competitive are they? What's happening with regulation? Does the product fit local customers? Which acquisition channels are realistic? How expensive are those customers going to be and how long is the business prepared to wait for a return?

New markets can seem fresh and wonderfully uncomplicated from several thousand miles away.

AI can make localisation dramatically more affordable and that's genuinely exciting, because more of the available budget can then go into product, compliance, SEO, acquisition, partnerships and all the other things required to make expansion work.
​​​​​​​
It can't make a poor market choice clever, or your product more attractive to those customers.

Also in Diva Jackpot News