id string | slug string | lang string | content_type string | doc_type string | category string | title string | summary string | text string | tags list | date timestamp[s] | url string | ai_generated bool |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
agentic-rag-confiavel-para-plataformas-enterprise:pt | agentic-rag-confiavel-para-plataformas-enterprise | pt | article | comparison | ai-agents | Agentic RAG na AWS: Comparativo de Arquiteturas para Ambientes Financeiros | Agentic RAG deixou de ser experimento de laboratório e passou a ser requisito de plataforma em ambientes financeiros que exigem rastreabilidade, controle de custo e latência previsível. Neste artigo, comparo quatro abordagens arquiteturais concretas na AWS, com trade-offs reais, números plausíveis e uma recomendação se... | Quando o Google Research publicou sua análise sobre Agentic RAG confiável para plataformas enterprise, o sinal que captei não foi sobre modelos, foi sobre orquestração, governança e onde a responsabilidade de controle deve residir na pilha. Em ambientes financeiros onde cada chamada de LLM precisa ser auditável, cada f... | [
"agentic-rag",
"bedrock",
"opensearch",
"eks",
"step-functions",
"financial-grade",
"governance",
"aws"
] | 2026-06-13T00:00:00 | https://fernando.moretes.com/blog/agentic-rag-confiavel-para-plataformas-enterprise?lang=pt | true |
agentic-rag-confiavel-para-plataformas-enterprise:en | agentic-rag-confiavel-para-plataformas-enterprise | en | article | comparison | ai-agents | Agentic RAG on AWS: Architecture Bake-Off for Financial-Grade Platforms | Agentic RAG has moved from lab experiment to platform requirement in financial environments that demand auditability, cost control, and predictable latency. In this article I compare four concrete architectural approaches on AWS, with real trade-offs, plausible numbers, and an unambiguous recommendation. | When Google Research published its analysis on reliable Agentic RAG for enterprise platforms, the signal I captured was not about models, it was about orchestration, governance, and where control responsibility should live in the stack. In financial environments where every LLM call needs to be auditable, every retriev... | [
"agentic-rag",
"bedrock",
"opensearch",
"eks",
"step-functions",
"financial-grade",
"governance",
"aws"
] | 2026-06-13T00:00:00 | https://fernando.moretes.com/blog/agentic-rag-confiavel-para-plataformas-enterprise?lang=en | true |
ai-bots-waf-e-monetizacao-de-trafego-com-seguranca:pt | ai-bots-waf-e-monetizacao-de-trafego-com-seguranca | pt | article | technology-review | security-resilience | AWS WAF e Monetização de Tráfego de Bots de IA: Análise Técnica | O AWS WAF ganhou capacidade nativa de identificar e rotear tráfego originado por bots de IA, uma mudança que transforma uma ferramenta defensiva em um ponto de controle de receita. Neste artigo, analiso o que a funcionalidade entrega de fato, onde ela falha e como integrá-la com segurança em arquiteturas financeiras. | Por anos, tratamos bots no WAF como ameaças a bloquear. A nova capacidade de monetização de tráfego de bots de IA no AWS WAF inverte essa lógica: em vez de descartar requisições de crawlers de LLMs e agentes de IA, você pode identificá-los, classificá-los e encaminhá-los para endpoints pagos, cobrar pelo acesso aos seu... | [
"aws-waf",
"ai-bots",
"security",
"finops",
"api-gateway",
"monetization",
"edge",
"zero-trust"
] | 2026-06-15T00:00:00 | https://fernando.moretes.com/blog/ai-bots-waf-e-monetizacao-de-trafego-com-seguranca?lang=pt | true |
ai-bots-waf-e-monetizacao-de-trafego-com-seguranca:en | ai-bots-waf-e-monetizacao-de-trafego-com-seguranca | en | article | technology-review | security-resilience | AWS WAF and AI Bot Traffic Monetization: A Technical Review | AWS WAF has gained native capability to identify and route AI bot traffic, a shift that turns a defensive tool into a revenue control point. In this article, I analyze what the feature actually delivers, where it falls short, and how to integrate it safely in financial-grade architectures. | For years, we treated bots in WAF as threats to block. The new AI bot traffic monetization capability in AWS WAF inverts that logic: instead of discarding requests from LLM crawlers and AI agents, you can identify them, classify them, and route them to paid endpoints, charging for access to your data rather than simply... | [
"aws-waf",
"ai-bots",
"security",
"finops",
"api-gateway",
"monetization",
"edge",
"zero-trust"
] | 2026-06-15T00:00:00 | https://fernando.moretes.com/blog/ai-bots-waf-e-monetizacao-de-trafego-com-seguranca?lang=en | true |
amazon-bedrock-agentcore-harness-da-ideia-ao-agente-de-producao-amazon-bedro:pt | amazon-bedrock-agentcore-harness-da-ideia-ao-agente-de-producao-amazon-bedro | pt | article | deep-dive | ai-agents | Amazon Bedrock AgentCore Harness: Da Ideia ao Agente de Produção | O AgentCore Harness chegou ao GA em junho de 2026 como uma abstração gerenciada que colapsa o plano de controle de agentes LLM em dois chamadas de API. Neste artigo, analiso como o harness funciona internamente, onde ele falha, e o que arquitetos de sistemas financeiros precisam entender antes de colocá-lo em produção. | O problema que o AgentCore Harness resolve não é inteligência, é encanamento. Qualquer equipe competente consegue fazer um agente LLM funcionar em um laptop em uma tarde. O que explode o cronograma é a camada de produção: isolamento de execução, gerenciamento de sessão, roteamento de ferramentas, armazenamento de crede... | [
"bedrock",
"agentcore",
"agentic-ai",
"aws",
"financial-grade",
"observability",
"security",
"llm-orchestration"
] | 2026-06-24T00:00:00 | https://fernando.moretes.com/blog/amazon-bedrock-agentcore-harness-da-ideia-ao-agente-de-producao-amazon-bedro?lang=pt | true |
amazon-bedrock-agentcore-harness-da-ideia-ao-agente-de-producao-amazon-bedro:en | amazon-bedrock-agentcore-harness-da-ideia-ao-agente-de-producao-amazon-bedro | en | article | deep-dive | ai-agents | Amazon Bedrock AgentCore Harness: From Idea to Production-Grade Agent | AgentCore Harness reached GA in June 2026 as a managed abstraction that collapses the LLM agent control plane into two API calls. In this article, I analyze how the harness works internally, where it fails, and what architects of financial-grade systems need to understand before putting it into production. | The problem AgentCore Harness solves is not intelligence, it is plumbing. Any competent team can get an LLM agent running on a laptop in an afternoon. What blows up the timeline is the production layer: execution isolation, session management, tool routing, credential storage, distributed tracing, and the multiplicatio... | [
"bedrock",
"agentcore",
"agentic-ai",
"aws",
"financial-grade",
"observability",
"security",
"llm-orchestration"
] | 2026-06-24T00:00:00 | https://fernando.moretes.com/blog/amazon-bedrock-agentcore-harness-da-ideia-ao-agente-de-producao-amazon-bedro?lang=en | true |
amazon-bedrock-agentcore-otimizacao-continua-de-agentes-em-producao-amazon-bedro:pt | amazon-bedrock-agentcore-otimizacao-continua-de-agentes-em-producao-amazon-bedro | pt | article | technology-review | ai-agents | Amazon Bedrock AgentCore: Otimização Contínua de Agentes em Produção | O Amazon Bedrock AgentCore introduz um loop de melhoria contínua que transforma traces de produção em diagnósticos acionáveis, recomendações fundamentadas em dados e validação estatística via testes A/B. Para arquitetos de sistemas financeiros e plataformas de alto risco, isso representa a primeira tentativa séria da A... | O problema mais perigoso em sistemas agênticos não é o agente que falha com um stack trace visível, é o agente que responde com confiança, parece funcionar nos dashboards, e silenciosamente entrega respostas erradas para centenas de usuários durante semanas. O Amazon Bedrock AgentCore, com suas novas capacidades de oti... | [
"bedrock",
"agentcore",
"agentic-ai",
"observability",
"mlops",
"aws",
"financial-grade",
"a-b-testing"
] | 2026-06-23T00:00:00 | https://fernando.moretes.com/blog/amazon-bedrock-agentcore-otimizacao-continua-de-agentes-em-producao-amazon-bedro?lang=pt | true |
amazon-bedrock-agentcore-otimizacao-continua-de-agentes-em-producao-amazon-bedro:en | amazon-bedrock-agentcore-otimizacao-continua-de-agentes-em-producao-amazon-bedro | en | article | technology-review | ai-agents | Amazon Bedrock AgentCore: Continuous Agent Optimization in Production | Amazon Bedrock AgentCore introduces a continuous improvement loop that turns production traces into actionable diagnostics, data-grounded recommendations, and statistical validation via A/B testing. For architects of financial systems and high-stakes platforms, this represents AWS's first serious attempt to close the g... | The most dangerous problem in agentic systems is not the agent that fails with a visible stack trace, it is the agent that responds confidently, looks fine on dashboards, and silently delivers wrong answers to hundreds of users for weeks. Amazon Bedrock AgentCore, with its new continuous optimization capabilities annou... | [
"bedrock",
"agentcore",
"agentic-ai",
"observability",
"mlops",
"aws",
"financial-grade",
"a-b-testing"
] | 2026-06-23T00:00:00 | https://fernando.moretes.com/blog/amazon-bedrock-agentcore-otimizacao-continua-de-agentes-em-producao-amazon-bedro?lang=en | true |
aml-alert-triage-amazon-quick-mcp-snowflake:pt | aml-alert-triage-amazon-quick-mcp-snowflake | pt | article | deep-dive | financial-systems | Triagem de Alertas AML com IA Governada: Arquitetura e Trade-offs | Automatizar a triagem de alertas de AML com IA generativa é tecnicamente viável, mas a distância entre um protótipo funcional e um sistema financeiro auditável é enorme. Neste artigo, analiso a arquitetura real por trás dessa automação, os mecanismos de orquestração, os pontos de falha silenciosos e as decisões de desi... | Triagem de alertas de Anti-Lavagem de Dinheiro (AML) é um dos problemas mais caros e menos glamourosos do setor financeiro: analistas humanos descartam manualmente entre 90% e 98% dos alertas gerados por sistemas de monitoramento de transações como NICE Actimize ou Oracle FCCM, a maioria falsos positivos. A promessa de... | [
"aml",
"financial-services",
"generative-ai",
"mcp",
"snowflake",
"bedrock",
"governance",
"compliance"
] | 2026-06-12T00:00:00 | https://fernando.moretes.com/blog/aml-alert-triage-amazon-quick-mcp-snowflake?lang=pt | true |
aml-alert-triage-amazon-quick-mcp-snowflake:en | aml-alert-triage-amazon-quick-mcp-snowflake | en | article | deep-dive | financial-systems | AML Alert Triage with Governed AI: Architecture and Trade-offs | Automating AML alert triage with generative AI is technically feasible, but the distance between a working prototype and an auditable financial system is vast. In this article, I analyze the real architecture behind this automation, the orchestration mechanisms, the silent failure points, and the design decisions that ... | Anti-Money Laundering (AML) alert triage is one of the most expensive and least glamorous problems in financial services: human analysts manually dismiss between 90% and 98% of alerts generated by transaction monitoring systems like NICE Actimize or Oracle FCCM, most of them false positives. The promise of using genera... | [
"aml",
"financial-services",
"generative-ai",
"mcp",
"snowflake",
"bedrock",
"governance",
"compliance"
] | 2026-06-12T00:00:00 | https://fernando.moretes.com/blog/aml-alert-triage-amazon-quick-mcp-snowflake?lang=en | true |
avaliacao-agentes-bedrock-agentcore-datasets:pt | avaliacao-agentes-bedrock-agentcore-datasets | pt | article | trend-briefing | ai-agents | Avaliação de Agentes como Disciplina de Engenharia | A avaliação de agentes de IA deixou de ser um exercício ad hoc de prompt engineering e tornou-se uma disciplina de engenharia com datasets versionados, quality gates automatizados e rastreabilidade de regressões. O Bedrock AgentCore materializa esse salto ao trazer infraestrutura gerenciada para o ciclo de vida de test... | Durante anos, a avaliação de sistemas de IA em produção foi tratada como um problema de ciência de dados, um notebook Jupyter rodado antes do deploy, métricas de acurácia calculadas offline, e a esperança de que o comportamento em produção refletisse o que foi medido no laboratório. Agentes de IA quebraram esse contrat... | [
"bedrock-agentcore",
"agent-evaluation",
"ai-testing",
"quality-gates",
"mlops",
"financial-grade",
"devops",
"observability"
] | 2026-06-06T00:00:00 | https://fernando.moretes.com/blog/avaliacao-agentes-bedrock-agentcore-datasets?lang=pt | true |
Architecture Decisions PT/EN
210 architecture documents, each in Brazilian Portuguese and English: 420 rows. It covers ADRs, design docs, postmortems, playbooks, teardowns and technical articles about AI agents, resilience, financial systems, FinOps and data platforms on AWS.
Portuguese technical material on these subjects is scarce, and pairs that cover the same document in both languages are scarcer. That pairing is the point of this dataset.
| Count | |
|---|---|
| Rows | 420 (210 pt, 210 en) |
| Unique documents | 210 (89 studies, 121 articles) |
| Search chunks | 9,741 |
| Most common study types | ADR 25, postmortem 20, playbook 13, design doc 12 |
Read this before using it
All of this content is AI-generated. It was written by language models following my editorial direction, voice guide and source curation, and it is published on fernando.moretes.com. Every row carries ai_generated: true.
- Postmortems, ADRs and teardowns are study material, not records of real incidents or decisions at any company, unless the document itself cites a public source.
- Limits, prices and service features change. Check the official documentation before relying on them.
- This is a snapshot taken on 2026-11-05. The site keeps changing and this file does not follow it automatically.
Fields
id, slug, lang, content_type (article or study), doc_type, category, title, summary, text, tags, date, url, ai_generated.
The same slug in both languages identifies a pair. Keep pairs in the same split, otherwise your evaluation leaks.
from datasets import load_dataset
ds = load_dataset("fernandofa42/architecture-decisions-pt-en", split="train")
adrs_pt = ds.filter(lambda r: r["doc_type"] == "adr" and r["lang"] == "pt")
search/ holds the excerpts and int8 embeddings used by the Architecture Search Space (paraphrase-multilingual-MiniLM-L12-v2, 384 dimensions).
License
CC BY 4.0. You can use, adapt and redistribute it, including commercially, as long as you credit "Fernando Azevedo, fernando.moretes.com" and indicate changes. The license covers the text written for this project; third-party material a document cites stays under its own terms.
Built by Fernando Azevedo, Specialist IT Architect focused on AI and AWS.
- Downloads last month
- 51