uMode raised R$2.7 million in 2023. The thesis was too dull for Silicon Valley: sell fashion AI to Brazilian retailers in reais. It now runs collection planning for more than three hundred brands, including Reserva and Vivara. The round was small enough to overlook; the companies that overlooked it still import their AI from California.

The consensus in 2023 held that Brazilian retailers would buy AI through the hyperscalers or not at all. A Linx survey reports 84 percent non-adoption; we argued yesterday that the figure measures the wrong thing. Reserva had by then already signed uMode, with prices in reais and terms Google Cloud would not write.

Smart Money Ventures and Investidores.vc led the round, targeting R$6.5 million in annual recurring revenue by year-end. Scale was not the thesis; the case study was.

The Reserva case study shows four times the product-development throughput at constant headcount, a quarter cut in styling follow-up and more than thirty percent fewer engineering errors once the platform displaced the spreadsheets. Three-month pilot; eighteen-month renewal; more than two hundred hours of customisation; more than a hundred and sixty more for ERP integration. The Google Cloud contract does not include those hours. A Brazilian vendor does.

uMode’s advantage was that it could be argued with in Portuguese.

A Brazilian brand does not need a forecast tuned on American denim cycles. It needs one that knows Black Friday in Brazil is a summer sale and that consumer payment runs to twelve interest-free instalments. uMode’s founders spent years inside Brazilian fashion before writing any code; the code arrived already Brazilian. The platform reads Portuguese social chatter for demand signals and ships a PLM that integrates with SAP and Lynx, the ERPs common in Brazilian fashion retail. That is the local-vendor advantage the hyperscalers cannot price.

The same pattern has played out elsewhere in LATAM across the last decade. Rappi wrote logistics for the cities Amazon could not afford, and Mercado Libre underwrote the consumer credit book Visa refused to open. Apparel lagged because the software runs deeper into the operation, the margins thinner, and because the US vendors large enough to serve a Reserva had reason to prioritise Nordstrom. uMode’s raise was the moment an AI-specific vendor crossed that same threshold in apparel.

Brazilian apparel is now large enough to fund its own SaaS stack. Reserva renewed an eighteen-month contract after a three-month pilot. Puket, Caedu and the rest of uMode’s three-hundred-brand roster use the same platform; uMode counts five thousand professionals on it and 542,000 machine-generated communications in 2024. None of them is writing dollar cheques to Google Cloud for a feature uMode already ships in reais. None is waiting for a US vendor to notice the market exists.

The story of apparel AI in 2026 is that it fragments by geography before it consolidates by capability. Silicon Valley cannot serve the Brazilian mid-market at the price the mid-market will pay; a São Paulo vendor can, and is. That is not a gap in the market. It is the market.