Speaking Every Customer’s Language: Why Localisation Is the Hidden Foundation of Agentic Commerce
For the last two decades, online retail has been a game of storefronts. You built a website, optimised it for search, drove traffic to it and worked hard to convert that traffic once it arrived. The shop window was yours and the customer came to you.
However, agentic commerce quietly rewrites that arrangement. As shoppers increasingly begin their journey inside an AI assistant (asking ChatGPT or Claude AI to compare options, recommend a product or simply buy the thing that best fits their need), discovery, comparison and checkout collapse into a single conversation. The customer no longer browses your storefront.
An AI surfaces, summarises and recommends your products on your behalf, often before the shopper has visited a single website.
Most of the conversation about this shift focuses on technical SEO, structured product data, schema mark-up and brand visibility in AI answers. All of that matters. But there is a layer underneath it that SME retailers rarely plan for until it becomes urgent: language. And not just language in the narrow sense of translation: tone of voice, cultural nuance and the visual assets that carry your brand across every market you sell into.
Agentic Commerce is Inherently Borderless
Here is the part that catches brands off guard. A traditional website has a natural gravitational pull towards its home market: your domain, your currency, your default language. An AI assistant has no such gravity. When a shopper in Mexico City, Riyadh or Warsaw asks an AI for a recommendation, the assistant answers in their language, drawing on whatever product information it can understand and trust.
That means your products are now being evaluated, described and recommended in markets you may never have deliberately targeted, and in languages your content was never written for. If your product information only exists in clean, confident English, you are effectively invisible or, worse, misrepresented the moment the conversation happens in another language.
For SME retailers, this is both a risk and an opportunity. The risk is being filtered out of AI recommendations in high-intent international conversations. The opportunity is that localisation is no longer a heavyweight, enterprise-only project. Done well, it becomes one of the cheapest ways to expand your addressable market without opening a single new warehouse.
Why Basic Translation Fails
The temptation is to run your product copy through a translation tool and move on. For a handful of low-stakes strings, fine. For a brand trying to win trust in an AI-mediated purchase, it breaks in three places, and each one maps to a layer you actually need to get right.
Language that’s accurate, not just correct
A dictionary lookup gets you words that are technically right but culturally deaf. Colours, idioms, humour, sizing conventions, even the way a benefit is framed all shift between markets; a line that lands in the UK can read as confusing or tone-deaf elsewhere. The fix isn’t translation but adaptation, what the industry calls transcreation, and it’s exactly what AI assistants reward, because they’re trained on how real people in that market actually speak.
Tone that still sounds like you
The wit, the warmth, the specific way you describe a product, the things that make your brand distinctive, rarely survive a word-for-word conversion. What comes out is forgettable, and in a market where an AI is comparing you on tone and clarity as much as price, generic is a quiet killer. Holding the line here means locking product names, taglines and do-not-translate terms up front in a brand glossary, so your voice carries instead of dissolving.
Visuals that travel with the words
Most localisation stops at the text and forgets how much retail meaning lives inside images: packaging, lifestyle shots, on-pack claims, promotional banners. Localise the copy but leave English packaging or an untranslated call to action, and the experience feels broken. Done right, the creative itself adapts (text fits, fonts hold, layouts stay intact, right-to-left languages like Arabic flip properly) without rebuilding every asset by hand, market by market, which is the slow, expensive grind that stops SME brands from going global in the first place.
When these three move together, something useful happens for agentic commerce specifically: your product information becomes consistent and trustworthy across every language
AI assistants favour brands whose data is clean, coherent and well-represented. Fragmented, half-translated, off-brand content is exactly the noise that gets a product passed over in an AI recommendation.
A Practical Guide for SME Brands
You do not need a global rollout to benefit from this. A pragmatic sequence works far better:
- Pick your highest-intent markets first. Look at where international demand is already leaking in (analytics, marketplace data, even where AI tools are surfacing you) and start there rather than trying to cover the world at once.
- Build a brand glossary before you scale. Lock product names, key claims and tone-of-voice rules up front. This is the single highest-leverage thing you can do to keep quality consistent as volume grows.
- Localise inside your existing tools and files. The goal is to adapt product listings, campaigns and creative directly within the systems and design files you already use (your store platform, your marketplaces, your artwork) rather than recreating everything per market.
- Use AI for the volume, keep humans on the nuance. Let automation absorb the repetitive load so your team’s judgement is spent where it counts: tone, cultural fit and the handful of decisions that genuinely shape the brand.
- Treat product data as a living asset. In an AI-mediated market, your structured, localised product information is what gets you recommended. It deserves the same care you give your storefront.
The Bigger Picture
For SME retailers in particular, this is encouraging news. You no longer need an enterprise budget and a year-long programme to compete internationally. You need clean product data, a clear brand voice, and a way to carry both (words and visuals) faithfully into every language your customers, and their AI assistants, happen to use.
The brands that internalise this early will quietly expand into markets their competitors never even realise they are losing. In the age of agentic commerce, speaking your customer’s language (properly, on-brand, and all the way through to the creative) is no longer a nice-to-have. It is the price of being recommended at all.
Frequently Asked Questions
The following FAQs will hopefully answer any immediate questions but please add a comment below if you have a more specific question.
What is Agentic Commerce?
Do I need to localise for every market?
What's the difference between translation and transcreation?
How does localisation influence whether AI assistants recommend a product?
Is localisation only realistic for large, enterprise-level brands?
What happens to product images and packaging when content is localised?
How are right-to-left languages like Arabic handled in localised creative?
I'm an SME business owner, is this relevant to me or just for bigger brands?
This article was contributed by Ashish Ranjan Jha, founder of Nativ (usenativ.com), an AI platform for content and visual asset localisation. Nativ helps brands adapt copy and creative across languages and markets, preserving brand voice and visual integrity, and is backed by a16z Speedrun and used by a wide range of brands from Sacheu Beauty to Ooredoo Fintech.
