Latin America's AI Commerce Gap:
75 Sites, One Painful Blind Spot
We scanned 75 of the most visited e-commerce sites across Brazil, Mexico, Chile, Argentina, Colombia and beyond: the stores that millions of LATAM shoppers rely on every day. The average overall readiness score was 54/100. The average transactability score was 29.4. AI agents can browse these stores. They cannot buy from them.
Latin America is now the world's fastest-growing retail e-commerce market. The region hit $191 billion in online retail in 2025, growing 12.2% year-on-year, and is projected at a CAGR of 9.43% through to 2029, second fastest globally. Brazil accounts for 45% of regional revenues.[1] Mexico's B2C e-commerce hit $38 billion in 2025, up 21% in a single year.[2] These are serious, scaled businesses, not underfunded operations with basic storefronts.
And yet: 76% of Brazilian businesses and 70% of Mexican businesses already use generative AI.[3] LATAM is not behind on AI adoption. It is behind on a very specific thing: the programmatic commerce infrastructure that lets AI agents act as buyers on a website. When we measured these 75 stores against that criteria (can an AI agent find your products, understand them, and complete a purchase without human intervention?) the picture was stark.
The single biggest gap is transactability: a score of 29.4 out of 100. AI agents can often find and read LATAM product pages. When they try to actually buy, the infrastructure simply is not there.
Where the scores break down
Every site we scan is measured against the Aidō DUT framework: three dimensions that together determine whether an AI agent can actually transact on a store, not just visit it.
The dimension breakdown reveals a structural imbalance. LATAM stores have invested in product content and page design. They have not invested in the programmatic commerce infrastructure that AI agents require.
Discoverability and understandability are in the moderate range. Product pages exist, HTML is parseable, and some schema markup is present. But transactability (the measure of whether an AI agent can programmatically add to cart, apply discounts, and complete checkout) sits at less than 30 out of 100. That is a structural problem, not a cosmetic one.
Score distribution
Overall scores ranged from 23 to 84. The top performer was a Chilean supermarket chain. The bottom scorers (including some of the region's largest retail brands) sat at or below 25. Seventeen sites scored zero entirely, blocked by bot protection before we could measure anything.
The sites that stood out
A handful of LATAM retailers are meaningfully ahead. Pet retail and beauty categories led the pack, likely because their product catalogues are naturally well-structured and their customers already expect click-and-collect and delivery infrastructure that translates well to programmatic commerce.
| Category | Country | Score | Bar |
|---|---|---|---|
| Supermarket chain | 🇨🇱 Chile | 84 | |
| Beauty & cosmetics retailer | 🇧🇷 Brazil | 80 | |
| Pet supplies retailer | 🇲🇽 Mexico | 78 | |
| Pet supplies retailer | 🇧🇷 Brazil | 76 |
At the other end, some of the most widely recognised names in LATAM retail scored in the low-to-mid twenties. These are not failing businesses. They are some of the most visited retail destinations in their respective markets. But their web infrastructure was built for human shoppers navigating browser sessions, not AI agents making programmatic requests.
| Category | Country | Score | Bar |
|---|---|---|---|
| Mass-market supermarket chain | 🇲🇽 Mexico | 23 | |
| Discount grocery chain | 🇲🇽 Mexico | 23 | |
| Regional food delivery platform | 🌎 Regional | 24 |
Markets covered
Our scan covered the primary e-commerce markets across the region, with the heaviest representation in Brazil and Mexico, the two largest markets by digital commerce volume.
The four gaps holding LATAM back
The leapfrog problem
The transactability number makes more sense once you know how LATAM actually shops. This region did not go from desktop to mobile. It went almost directly from cash to smartphones, bypassing the desktop e-commerce era that set the template for web commerce in Europe and North America.
Smartphones account for 83% of all mobile connections in Latin America in 2025, with Brazil at 88% and rising.[4] But device penetration is only part of the story. The more important shift is where commerce actually happens: not on web storefronts, but in WhatsApp. Latin America's conversational commerce market (buying and selling inside chat threads) is a $18.2 billion market in 2025, growing at 35% year-on-year, with 72% of that volume flowing through WhatsApp. In Brazil alone, 78% of businesses make active sales via WhatsApp. AI-assisted WhatsApp sales convert at 45 to 55%, against 1.5 to 2.1% for a standard mobile website, a difference of 13 to 19 times.[5]
Mexico added QR code payments inside WhatsApp Business in 2025.[6] Brazil's Pix payment network integrates natively into WhatsApp chats. Real commerce infrastructure, built to work at scale. The problem is it runs through chat threads, not APIs a shopping agent can reach.
Aidō's read on this: LATAM largely skipped the programmatic web commerce layer that agentic shopping infrastructure runs on. It went from cash to mobile and conversational channels without stopping to build the API-first web infrastructure in between. Cash's share of in-store payments fell from 67% in 2014 to 25% by 2025,[7] while Pix in Brazil and SPEI in Mexico absorbed much of that shift digitally, through app and chat interfaces rather than web checkout flows. In our view, the region is not behind on AI. It is ahead of it in conversational form. The website APIs that AI agents need to transact were simply never a priority to build.
Why LATAM is particularly exposed
Most of what we scanned was built during the e-commerce boom of the 2010s and early 2020s, when mobile-first, app-centric, marketplace commerce was what mattered. That was the right call at the time. The problem is that the same choices (prioritising mobile UX, WhatsApp flows, local payment methods over browser-accessible APIs) now create exactly the friction that stops AI agents cold.
Brazil's CPF requirement, Mexico's RFC validation, and Argentina's multi-currency complexity all add layers to checkout that are difficult to navigate programmatically.[8] These are not technical failures. They reflect real regulatory and economic realities. But they do mean that LATAM retailers face additional work before their stores are agent-ready, compared to markets where checkout is simpler.
Note on language: This report uses "LATAM" as shorthand. Our scan included Portuguese-language Brazilian stores alongside Spanish-language markets. Brazil dominated the top scores, driven mainly by beauty and pet retail. The challenges are regional; the fixes are not.
Aidō's view: mobile-first becomes agent-first
LATAM did not fail to adopt modern commerce. It adopted a different version: phone-first, WhatsApp-native, built for the infrastructure and consumer habits of the last decade. That was the right call.
What comes next probably is not a reckoning. The conversational instinct that LATAM consumers already have (tell an AI what you want, get a price, pay in the chat) maps almost perfectly onto how agentic shopping works. The difference is scale and autonomy. A WhatsApp sales agent handles one conversation at a time. An AI shopping agent browses, compares, negotiates, and buys across multiple stores simultaneously, without any human in the loop.
The question for LATAM retailers is not whether to participate in that future. Their customers are already demanding it. The question is whether their store infrastructure will be on the other side of that interaction as a destination, or invisible to it.
The LATAM retailers who move earliest on three things will capture most of the agent-driven purchase volume: structured product APIs that agents can query without a browser; ACP coverage so agents know what the store can do; and a deliberate policy of allowing verified agent traffic rather than blocking it. The WhatsApp conversion rates (13 to 19 times higher than mobile web) already prove that when the friction goes, LATAM consumers buy. The question is who will be there to take the order.
What the leaders are doing differently
The sites that scored in the 70s and above have a few things in common. They tend to operate in categories with naturally structured product data: pet food, health and beauty, grocery. A bag of dog food has a weight, a brand, a SKU, and a price that doesn't change hourly. Schema markup is easy when the data is already clean.
They have also tended to invest in headless or composable commerce architectures that expose APIs as a natural consequence of how they are built, rather than as a bolt-on. The top-scoring supermarket in our dataset allows programmatic product queries even without explicit ACP support.
For everyone else: structured data, API-first infrastructure, and a decision to treat verified agent traffic as customers rather than bots.
Want to know where your store sits?
We scan e-commerce sites across LATAM and globally, and tell you exactly what an AI agent can and cannot do on them. Drop us a line and we'll run the numbers.
Get in touch →Scan methodology: 75 LATAM e-commerce sites scanned by Aidō Lighthouse, April 2026. 58 sites returned scoreable results; 17 were blocked by bot protection before any data could be collected. Overall score is a weighted composite of discoverability (35%), understandability (40%), and transactability (25%) using the Aidō DUT framework. Score distribution bands are approximate. Countries included: Brazil, Mexico, Chile, Argentina, Colombia, Peru, Uruguay and regional cross-border platforms. All scores reflect site state at time of scan. The editorial interpretation of LATAM commerce patterns as they relate to agentic commerce readiness is Aidō's own view, informed by the scan data and the sources below.
- LATAM e-commerce market size ($191B), 12.2% year-on-year growth, 9.43% CAGR to 2029, and Brazil's 45% revenue share: eMarketer, Latin America Ecommerce Forecast 2025.
- Mexico B2C e-commerce ($38B, 21% growth): Antom / PaymentsCMI, LATAM Digital Commerce Report 2025.
- AI adoption rates (76% of Brazilian businesses, 70% of Mexican businesses): Citrusbug / EComposer, AI Adoption in Latin American E-Commerce, 2025.
- Smartphone penetration (83% of mobile connections in Latin America, Brazil at 88%): GSMA Intelligence, Mobile Economy Latin America 2025.
- Conversational commerce market size ($18.2B, 35% YoY growth), WhatsApp's 72% share of that volume, 78% of Brazilian businesses selling via WhatsApp, and AI-assisted conversion rates (45–55% vs 1.5–2.1% on mobile web): AuroraInbox, LATAM Conversational Commerce Report 2025; corroborated by Statista Digital Market Outlook 2025.
- WhatsApp Business QR code payment integration in Mexico: Meta / WhatsApp Business product announcements, 2025. Brazil Pix native integration in WhatsApp: Banco Central do Brasil and Meta joint communications, 2025.
- Cash's share of in-store payments falling from 67% (2014) to 25% (2025): PYMNTS, Digital Payments in Latin America 2025.
- Analysis of CPF, RFC, and multi-currency checkout friction for AI agents: Aidō Lighthouse scan data, April 2026, corroborated by Brazil Receita Federal and Mexico SAT public documentation on electronic tax invoice (nota fiscal / factura) requirements.