# Bex S.A. -- bexbr.com # Full site content in English for LLM consumption. # Site language: Portuguese (pt-BR) -- content below is a faithful translation. # Summary: /llms.txt # Site: https://bexbr.com --- ## Home -- bexbr.com ### The infrastructure that makes autonomous mobility possible in Brazil. We believe computing is leaving screens to inhabit the physical world. Autonomous vehicles are the most concrete expression of this transformation -- and Brazil cannot be just a spectator. We build on three fronts. 1. **Data** -- Continuous video and telemetry collection from Brazilian traffic at scale, with Bex Cam and Malha Bex, generating the Bex Dataset. 2. **Protocols** -- Vehicle-to-city communication standards adapted to our roads, through Bex V2X. 3. **Models** -- The Brazilian autonomous driving stack -- perception, decision, and control trained on our data. What Autopilot is for Tesla, Bex Pilot is for Brazil. What the infrastructure enables: 4. **Bex Platform** -- Complete automotive platform for autonomous vehicles -- data, protocols, and models in a single system. For partner manufacturers or proprietary vehicles. 5. **Bex AI** -- The conversational interface of this intelligence. A copilot inside the vehicle -- navigation, alerts, voice. 6. **Bex Ride** -- Autonomous robotaxis on Brazilian streets. Individual transport at scale -- no driver, no wait. --- ## Why we exist -- bexbr.com/missao ### Our mission is to make autonomous mobility a reality in Brazil. For a vehicle to drive itself on Brazilian streets, it is not enough to bring technology from abroad. No model trained in São Francisco understands a motorcycle splitting lanes at 80km/h. No European protocol was designed for our road signs. No dataset includes our highways. The invisible infrastructure is missing: real data from our traffic, communication protocols adapted to our roads, and AI models trained on our reality. Without this, autonomous mobility in Brazil does not exist. That is what Bex builds. Not an app. Not a feature. The foundation -- the hardest and most fundamental layer, upon which everything that comes after will be built. A country that gridlocks in traffic, that depends entirely on highways, and that lacks accessible transport at scale is the scenario that has the most to gain from this transformation. But you do not import the future. You build it. --- ### What changes when it works. In 2035, an elderly person in Sao Paulo calls a vehicle at three in the morning for an emergency -- without depending on anyone. A child goes to school while parents work. A truck travels highways at night, without a driver, delivering what the country needs. This is not science fiction. It is the natural consequence of solving an engineering problem: making machines understand the world around them and move safely within it. Autonomous mobility will reorganize cities, logistics, and routines more profoundly than the automobile reorganized the 20th century. It is one of the greatest economic and social transformations of the century. --- ### Every technological revolution is structured in layers. Infrastructure, platforms, applications, transactions. Whoever builds the infrastructure captures disproportionate and lasting value -- because everything that comes after depends on it. Think of the AI revolution. NVIDIA GPUs and AWS servers are the infrastructure. Claude, ChatGPT, Gemini are the platforms. Cursor, Perplexity are the applications. Who captured the most value? NVIDIA. Not by accident -- by position. | Infrastructure | Platforms | Applications | Transactions | |---------------|----------------|---------------------|-------------| | NVIDIA / AWS | Claude / GPT | Cursor / Perplexity | -- | | hardest | | | easiest | | most defensible | | | most replaceable | The same logic repeats in autonomous mobility. The infrastructure is the data, the protocols, the models. Bex operates in the first layer. By choice. --- ## Why Brazil -- bexbr.com/brasil ### Why Brazil needs to build its own foundations. The most advanced autonomous systems in the world were built with data from São Francisco, Phoenix, Shanghai, Seoul. Brazilian streets are a completely different domain. The motorcycle weaving between cars. The pedestrian crossing outside the crosswalk. The pothole in the middle of the road. No model trained outside Brazil will operate here safely without data from here. This is not a technical detail. It is a structural barrier. --- ### Brazil arrived late to the last revolutions. There is still time in this one. Brazil is the 2nd largest market for Uber and WhatsApp in the world. We consume technology at massive scale -- but barely produce the infrastructure that sustains it. In the internet, in cloud computing, in semiconductors -- Brazil was not at the table when the foundations were built. Autonomous mobility is still at the beginning. The infrastructure does not yet exist -- neither abroad nor in Brazil. It is one of the rare windows where arriving now still means arriving at the foundation. --- ### Mobility infrastructure is sovereignty infrastructure. The data that teaches vehicles to drive, the protocols that connect cars to the city, the models embedded in navigation systems -- whoever defines these standards defines how the country moves. > Either Brazil builds its own autonomous mobility infrastructure, or > imports it from outside and accepts its premises, limitations, and > dependencies. --- ## Products -- bexbr.com/produtos ### Data, protocols, models. Piece by piece. Each layer enables the next. **Data:** - Bex Cam -- video and telemetry collection device - Malha Bex -- mobile camera network in real vehicles - Bex Dataset -- first Brazilian traffic dataset for autonomous AI **Protocols:** - Bex V2X -- vehicle-to-infrastructure communication protocol **Models:** - Bex Pilot -- Brazilian autonomous driving stack (perception, decision, control) --- ## Bex Pilot -- bexbr.com/pilot The autonomous driving stack for the Brazilian context. Perception, decision, and control -- trained on Bex Cam data, adapted to real conditions on our roads. Competitors trained elsewhere: - Tesla Autopilot/FSD -- closed stack, American data, Tesla-only - Comma OpenPilot -- open source, retrofit, predominantly American data - Mobileye SuperVision -- sold to manufacturers, European/American data - Huawei ADS -- complete stack for Chinese manufacturers, Chinese data Bex Pilot fills the space none of these cover: a stack trained on the Brazilian context. --- ## About -- bexbr.com/sobre ### We build what seems too big. Nanpos Ventures is not a fund. We do not invest in other people's companies -- we build our own. Each subsidiary is a deliberate bet on a difficult, long-term problem that others avoid. **Subsidiaries:** - Bex S.A. -- Brazilian Autonomous Mobility Infrastructure - Findera -- Brazilian Local Data Infrastructure **Strategic alliances:** - Lecar S.A. -- Brazilian Electric Vehicle Industry - Ligeira S.A. -- Brazilian Electric Motorcycle Industry --- ## Company info | Field | Value | |---------------|-----------------------------------------------------------| | Legal name | Bex S.A. Infraestrutura Brasileira de Mobilidade Autonoma | | Group | Nanpos Ventures Participacoes e Investimentos LTDA | | Headquarters | Sao Paulo, Brazil | | Founder | Matheus Matos | | Domain | bexbr.com | | Founded | March 2026 | ## Contact -- bexbr.com/contato - Commercial partnerships: comercial@nanpos.com - Investors: matheus@nanpos.com - Careers: carreiras@nanpos.com - Press: imprensa@nanpos.com ## INPI Trademark Registrations -- 03/30/2026 | Application | Mark | Class | Type | |-------------|------------|-------|-----------| | 943211085 | BEX | 42 | Word mark | | 943211093 | BEX | 9 | Word mark | | 943211123 | BEX + logo | 42 | Combined | | 943211131 | BEX + logo | 9 | Combined | | 943211140 | NANPOS | 35 | Word mark |