Amazon Web Services brings Amazon Bedrock, a managed genAI service, to Brazil

By Amazon Web Services (AWS)

This is a machine translation of the original press release issued in Portuguese.

SÃO PAULO, JUNE 7, 2024 – Amazon Web Services, Inc. (AWS) announced today, June 7, the availability of Amazon Bedrock in the South America (São Paulo) cloud region, in Brazil. Fully managed, the service offers powerful capabilities for building and scaling generative artificial intelligence (AI) applications and makes it easy for customers to find and access high-performance foundational models for their specific use case.

Amazon Bedrock launched globally in select regions in 2023 and is now coming to Brazil, enabling customers across the country, including public sector organizations and companies in regulated industries, to innovate with a range of options for running and storing their AI applications generative.

“Generative AI is poised to be the most transformative technology of our time, and we are cutting through the noise to make it easy, practical and cost-effective for customers in Brazil,” says Cleber Morais, general director of Sales at Amazon Web Services in Latin America. “AWS is committed to helping customers accelerate their generative AI journey, from experimentation to production, efficiently and securely. Our customers value flexibility of choice: some prefer to build from scratch, while others opt for models adapted to specific cases.”

Companies across Brazil are using generative AI for a variety of purposes, including increasing productivity, creating innovative experiences for users and even reinventing the work environment. Amazon Bedrock is a fully managed service that offers customers a selection of high-performance foundation models (FMs) and robust model customization capabilities, built with enterprise-grade security and privacy. Additionally, AWS provides customers with the tools, resources, and training needed to drive responsible and safe innovation in generative AI.

"We're excited to help customers accelerate innovation with generative AI across industries with Amazon Bedrock. In the financial services sector, for example, investment brokers can use generative AI to continually optimize their strategies based on portfolio performance in real time and under market conditions. In the healthcare sector, hospitals can use virtual care assistants to alleviate resource limitations, using conversational language to answer patients' questions”, explains Morais.

Deploying AI workloads closer to end users will help customers with low latency needs, especially important for generative AI applications. Low latency provides faster processing and response times essential for AI tasks such as real-time content generation, interactive user experiences, and real-time conversational insights. Starting today, customers can access the latest generation of templates that include Amazon Titan (Text Lite, Text Express, Multi-modal Embeddings, Titan Text Embeddings v2), Anthropic (Claude 3 Haiku, Claude 3 Sonnet), Llama (Llama 3 8B, Llama 3 70B) and Mistral (Mistral Large, Mistral 7B, Mistral 8x7B).

AWS Customers and Partners Innovating with Generative AI

Bravante RCC , a robotics company specializing in the shipping, port, and oil and gas industries, uses AWS technology to locate decommissioned oil pipelines. Generative AI document analysis helps determine the exact location of pipelines that need to be removed, a requirement of the National Petroleum Agency (ANP), reducing the burden on surveying engineers. Another important focus is the reduction of planning errors in decommissioning which, in addition to social and environmental impacts, can cause losses of more than US$250,000 each. Partner Flexa Cloud created a customized solution with Amazon Bedrock, using Anthropic Claude 2's LLM, among other AWS products.

C6 Bank, one of the 345 most promising startups in the world, according to North American consultancy CB Insights, is developing two solutions based on generative AI and analytics with AWS. In practice, this tool will warn you if there is anything non-standard during application development and, if so, how the code should be changed. The second solution is a robot that scans the database of information about the bank's products and services and makes suggested responses. The attendant has the option of sending the answer in full, using only some excerpts from the suggested text or discarding the suggestion. More elaborate than a bot based on a simple tree of words, the resource interprets the customer's doubt, cross-references the information with the knowledge base and formulates answers based on its findings, simulating human writing.

Gimba, a leading retail provider of essential materials and supplies, has replaced the manual process of inserting and updating products in the catalog with a generative AI tool. Technology is currently responsible for the inclusion of around 300 new products/month, reducing registration time by more than 80%.

The InovaHC Lab, opened in 2024 within Hospital das Clínicas, is an interdisciplinary space that brings together researchers, students and entrepreneurs to explore applications, develop solutions and research. The objective of the Gen AI laboratory is to create applications that promote greater efficiency for public health diagnostic processes and transform the way radiological reports are created. Named GAL (Adaptive Report Generator), the tool allows the automatic structuring of radiological reports, taking into account the patient's history. This is expected to reduce the time spent by the radiologist in accessing the information necessary to analyze the images, bringing more efficiency to the diagnosis delivery process.

PagBank is a complete digital bank in financial services and payment methods. In order to speed up customer service, increase productivity and improve service quality, PagBank decided to test an artificial intelligence (AI) solution for cases of changing payment machines, one of the main reasons customers contact us. Using Amazon Rekognition to identify the serial number of machines through photos, in one month the bank reduced the average service time and the volume of recurring contacts by 85%. Amazon Bedrock will be used to answer out-of-scope questions based on frequently asked questions and answers. "Implementing artificial intelligence solutions in our customer service has brought a significant gain in innovation and agility. With AWS, we were able to develop a robust and effective solution, and the integration of advanced technologies allows us to not only automate processes, but also improve accuracy and the speed of service”, says Cesar Leite, CTO at PagBank.

SOMOS Educação, one of the main basic education companies in Brazil, with more than 2 million students, recently launched Plurall AI, a generative AI assistant to assist teachers and students in lesson planning. An integral part of the Plurall AI suite, the intelligent assistant 'Plu' offers personalized features including lesson plans, direct book queries, content summaries, test creation and activity generation, reliably facilitating the educational process. The first case of an educational institution using Amazon Bedrock to adopt a generative AI solution in Brazil, the project uses LLM Claude, from Anthropic, and has the support of Accenture. “Artificial intelligence comes as a powerful tool for personalizing learning and adapting it to individual needs. With a focus on ambidextrous vision and constant innovation, Plurall AI, developed in partnership with AWS, is our great ally in this project. We already have the pilot project with partner schools underway, involving approximately 30 teachers and 400 students. The solution will be available to schools when back to school in 2025”, comments Guilherme Mélega, CEO of SOMOS Educação.

Support for training in generative AI

AWS continues to invest in training local developers and the next generation of IT leaders through training and certification programs. According to the AWS survey “Accelerating AI Skills: Preparing the Workforce for the Jobs of the Future,” conducted in partnership with the Access Partnership, 84% of employees in Brazil indicated an interest in developing AI skills to accelerate their careers and 73% of employers said hiring people skilled in Artificial Intelligence (AI) is a priority.

AWS is committed to providing free cloud computing skills training to 29 million people worldwide by 2025 and has already helped more than 21 million people acquire cloud skills in free courses, with more than 500 online courses available in more than 15 languages. Additionally, it recently announced a free and low-cost Amazon Bedrock Getting Started training series to help people understand, implement, and get started with generative AI.

“We are empowering organizations across all industries with the power of AI, investing in skills-enhancing initiatives, strengthening support programs for developers and startups, and expanding access to the latest generative AI services, including Amazon Bedrock, for our customers in Brazil”, adds Morais.

In November 2023, Amazon launched AI Ready, which offers a set of free courses and training in AI and generative AI, aligned to technical and non-technical roles, so that anyone can develop artificial intelligence skills. There is also new exclusive training for Brazilians, carried out in partnership with Trybe, the AWS Generative AI scholarship program, with the Udacity learning platform, and 8 free courses in AI and generative AI to support professionals looking to improve skills.

AWS has a long-term commitment to customers and local communities in Brazil, and maintains initiatives that play an important role in promoting economic development in the region. Since arriving in the country in 2011, AWS has invested more than R$19 billion to build, connect, operate and maintain its data centers. Since 2017, AWS has trained more than 800,000 Brazilians in cloud computing skills.

About Amazon Web Services

Since 2006, Amazon Web Services has been the world's most comprehensive and widely adopted cloud offering. AWS has been continually expanding its services to support virtually any cloud workload and now has more than 240 end-to-end services for compute, storage, databases, networking, analytics, machine learning and artificial intelligence (AI), Internet of Things (IoT), mobile, security, hybrid, media, development, deployment and management of media and applications from 105 Availability Zones across 33 geographic regions, with plans announced for an additional 18 Availability Zones and six additional regions across AWS in Malaysia, Mexico, New Zealand, Saudi Arabia, Thailand, and the AWS European Sovereign Cloud. Millions of customers, including the world's fastest-growing startups, largest enterprises, and leading government agencies, trust AWS to power their infrastructure, become more agile, and reduce costs. To learn more about AWS, visit aws.amazon.com.