<?xml version="1.0" encoding="UTF-8"?><rss version="2.0"><channel><title>guics</title><description>Texts by Guilherme da Cruz Silveira.</description><link>https://guics.me/en/</link><language>en-US</language><item><title>An AI-resolved request only counts if the customer doesn&#x27;t come back</title><link>https://guics.me/en/artigos/centauro-diz-que-ia-fecha-mais-de-60-dos-chamados-de-pos-venda-no-app/</link><guid isPermaLink="true">https://guics.me/en/artigos/centauro-diz-que-ia-fecha-mais-de-60-dos-chamados-de-pos-venda-no-app/</guid><description>The measurement applies only to the app, between July and August 2026, and the numbers come from the company itself. Even so, the case helps to think about the extra staff the sales peak requires and how to know whether the agent actually resolved the request.</description><pubDate>Thu, 12 Nov 2026 15:00:00 GMT</pubDate><author>Guilherme da Cruz Silveira</author></item><item><title>Early in companies&#x27; use of AI, it is possible to choose where to cut cost</title><link>https://guics.me/en/artigos/ainda-no-comeco-do-uso-de-ia-nas-empresas-da-para-escolher-onde-reduzir-custo/</link><guid isPermaLink="true">https://guics.me/en/artigos/ainda-no-comeco-do-uso-de-ia-nas-empresas-da-para-escolher-onde-reduzir-custo/</guid><description>Most companies still have their use of artificial intelligence at an early stage or in pilot projects. That is what the Folha coverage of the Amcham Brasil and Humanizadas survey shows. For those who lead an operation, this stage is the window to choose a task that cuts cost, shortens delivery time and reduces error or rework.</description><pubDate>Wed, 11 Nov 2026 15:00:00 GMT</pubDate><author>Guilherme da Cruz Silveira</author></item><item><title>Why I chose Grok Bot&#x27;s ecosystem of agents</title><link>https://guics.me/en/artigos/o-grok-bot-me-serve-porque-eu-monto-um-time/</link><guid isPermaLink="true">https://guics.me/en/artigos/o-grok-bot-me-serve-porque-eu-monto-um-time/</guid><description>I chose Grok Bot because I build a team of specialists, each one in its own role. In ChatGPT you can only have one dot, and Claude&#x27;s Duet also comes down to a single agent. I don&#x27;t spend my time correcting one assistant.</description><pubDate>Tue, 10 Nov 2026 15:00:00 GMT</pubDate><author>Guilherme da Cruz Silveira</author></item><item><title>Today is the best opportunity for AI in small and midsize companies</title><link>https://guics.me/en/artigos/idc-e-dell-57-das-empresas-ja-escalaram-ia/</link><guid isPermaLink="true">https://guics.me/en/artigos/idc-e-dell-57-das-empresas-ja-escalaram-ia/</guid><description>IDC, commissioned by Dell, surveyed 250 organizations in Brazil: 57% have moved past testing and scaled AI in their operations. Among small and midsize businesses, only 42% use the technology across several areas, against 72% of large companies. That gap, today, is the best opportunity.</description><pubDate>Tue, 10 Nov 2026 15:00:00 GMT</pubDate><author>Guilherme da Cruz Silveira</author></item><item><title>It sold, but the margin came up short</title><link>https://guics.me/en/artigos/vendeu-mas-a-margem-ficou-curta/</link><guid isPermaLink="true">https://guics.me/en/artigos/vendeu-mas-a-margem-ficou-curta/</guid><description>Márcio Nunes, from Asus, said at a retail event that maybe the best use of AI is helping a store not lose money. Mercado &amp; Consumo reports it. I&#x27;d start with the order that sold, but where the margin came up short.</description><pubDate>Mon, 09 Nov 2026 15:00:00 GMT</pubDate><author>Guilherme da Cruz Silveira</author></item><item><title>Snowflake announces weekly sales analyses with CoWork</title><link>https://guics.me/en/artigos/snowflake-anuncia-analises-semanais-de-vendas-com-cowork/</link><guid isPermaLink="true">https://guics.me/en/artigos/snowflake-anuncia-analises-semanais-de-vendas-com-cowork/</guid><description>With Snowflake CoWork, a sales leader can schedule analyses and get reports by email. Each run asks the question again with current data and…</description><pubDate>Mon, 09 Nov 2026 15:00:00 GMT</pubDate><author>Guilherme da Cruz Silveira</author></item><item><title>The agent can put the quote together. Who approves sending it?</title><link>https://guics.me/en/artigos/o-agente-pode-fazer-o-orcamento-quem-libera-o-envio/</link><guid isPermaLink="true">https://guics.me/en/artigos/o-agente-pode-fazer-o-orcamento-quem-libera-o-envio/</guid><description>Fleury says it uses an AI agent to calculate project costs, with the math available for review, according to Xataka. I see a good application there: letting the program run a whole step, with a clear limit on what it can do next.</description><pubDate>Sun, 08 Nov 2026 15:00:00 GMT</pubDate><author>Guilherme da Cruz Silveira</author></item><item><title>The end of the prompt: how to get the best answers out of AI</title><link>https://guics.me/en/artigos/o-prompt-nao-acabou/</link><guid isPermaLink="true">https://guics.me/en/artigos/o-prompt-nao-acabou/</guid><description>The best answer starts before the question. How to organize context, sources, skills and tools to work better with AI agents.</description><pubDate>Sun, 08 Nov 2026 15:00:00 GMT</pubDate><author>Guilherme Silveira</author></item><item><title>The customer&#x27;s history has to reach whoever is serving them</title><link>https://guics.me/en/artigos/o-historico-do-cliente-precisa-chegar-a-quem-atende/</link><guid isPermaLink="true">https://guics.me/en/artigos/o-historico-do-cliente-precisa-chegar-a-quem-atende/</guid><description>AI can help customer service resolve requests with the customer&#x27;s history at hand. It&#x27;s worth starting with that access. A survey by CS Academy with Zoho, reported by IT Forum, points to the lack of integration between systems as the most cited barrier to expanding the use of AI.</description><pubDate>Sat, 07 Nov 2026 15:00:00 GMT</pubDate><author>Guilherme Silveira</author></item><item><title>More agents, the same bad service: what the new SAC decree leaves out</title><link>https://guics.me/en/artigos/decreto-do-sac-atendimento/</link><guid isPermaLink="true">https://guics.me/en/artigos/decreto-do-sac-atendimento/</guid><description>Decree 13,135 is political. It requires people on duty, but it does nothing for the people who serve customers. At no point does the text mention education, qualification or training for the agent. That reinforces the thesis: it&#x27;s a populist add-on to keep a whole class employed, even with bad results. The law hangs more people on the sector without service improving, and the customer is still badly served.</description><pubDate>Fri, 06 Nov 2026 15:00:00 GMT</pubDate><author>Guilherme Silveira</author></item><item><title>AI that already delivers results; watch the data you send to the cloud</title><link>https://guics.me/en/artigos/a-ia-que-ja-da-resultado-cuidado-com-os-dados-enviados-a-nuvem/</link><guid isPermaLink="true">https://guics.me/en/artigos/a-ia-que-ja-da-resultado-cuidado-com-os-dados-enviados-a-nuvem/</guid><description>An IDC survey for Dell, published by PEGN, heard from 125 Brazilian companies with 99 to 499 employees: 46% already measure productivity gains with AI. A good sign, but the gain from the test is worth little if the company doesn&#x27;t know where the data lives, how much the cloud will charge and how to connect AI to the system it already uses.</description><pubDate>Fri, 06 Nov 2026 15:00:00 GMT</pubDate><author>Guilherme da Cruz Silveira</author></item><item><title>Anthropic tests a Claude that answers questions about account and card spending</title><link>https://guics.me/en/artigos/anthropic-testa-claude-que-responde-sobre-gastos-em-contas-e-cartoes/</link><guid isPermaLink="true">https://guics.me/en/artigos/anthropic-testa-claude-que-responde-sobre-gastos-em-contas-e-cartoes/</guid><description>Anthropic started testing in the United States a tool that connects bank accounts and credit cards to Claude, its AI assistant, to answer…</description><pubDate>Fri, 06 Nov 2026 15:00:00 GMT</pubDate><author>Guilherme da Cruz Silveira</author></item><item><title>With a clear rule, the agent checks and the process moves</title><link>https://guics.me/en/artigos/com-a-regra-clara-o-agente-confere-e-o-processo-anda/</link><guid isPermaLink="true">https://guics.me/en/artigos/com-a-regra-clara-o-agente-confere-e-o-processo-anda/</guid><description>At the Transamerica Expo Center, in São Paulo, the panel &quot;IA &amp; Agentes e o Seguro&quot; (AI &amp; Agents and Insurance) at CQCS Inovação 2026 brought together executives from Intelidata, Air Company and AXA in Brazil. Fabio Pereira, from Air Company, talked about agents that check a series of items and, if everything is within the rule, release steps like automatic payment. In practice, when the rule is clear, the step doesn&#x27;t have to wait for someone to look again.</description><pubDate>Fri, 06 Nov 2026 15:00:00 GMT</pubDate><author>Guilherme da Cruz Silveira</author></item><item><title>An order that arrives on WhatsApp doesn&#x27;t have to get lost anymore</title><link>https://guics.me/en/artigos/pedido-que-chega-no-whatsapp-nao-precisa-mais-se-perder/</link><guid isPermaLink="true">https://guics.me/en/artigos/pedido-que-chega-no-whatsapp-nao-precisa-mais-se-perder/</guid><description>Cluster, a marketing and technology company, gathered in one place what it knows about each client and each area and put AI agents to work in client service, creative, social media, ads and technology. Since October, these agents run on the company&#x27;s portal. For whoever runs the operation, AI starts to really help when it knows how the house works and turns a loose message into a logged request, with the name of whoever is going to do it.</description><pubDate>Fri, 06 Nov 2026 15:00:00 GMT</pubDate><author>Guilherme da Cruz Silveira</author></item><item><title>ANPD sandbox: Trajetto reaches 86 points</title><link>https://guics.me/en/artigos/sandbox-da-anpd-trajetto-chega-a-86-pontos-operacao-regular-exige-comprovacao/</link><guid isPermaLink="true">https://guics.me/en/artigos/sandbox-da-anpd-trajetto-chega-a-86-pontos-operacao-regular-exige-comprovacao/</guid><description>ANPD published the reports for cycles 4, 5 and 6 of its AI sandbox, with different results for three solutions. For Trajetto, the ethics and…</description><pubDate>Thu, 05 Nov 2026 15:00:00 GMT</pubDate><author>Guilherme da Cruz Silveira</author></item><item><title>Copastur expects to save 5,270 hours a month: the gain is in the dull routine</title><link>https://guics.me/en/artigos/copastur-estima-poupar-5-270-horas-por-mes-o-ganho-esta-na-rotina-chata/</link><guid isPermaLink="true">https://guics.me/en/artigos/copastur-estima-poupar-5-270-horas-por-mes-o-ganho-esta-na-rotina-chata/</guid><description>Copastur, a corporate travel agency, expects to save 5,270 work hours a month with AI agents, according to PANROTAS. We see a lot of companies…</description><pubDate>Wed, 04 Nov 2026 15:00:00 GMT</pubDate><author>Guilherme da Cruz Silveira</author></item><item><title>Productivity is what companies gain most from AI</title><link>https://guics.me/en/artigos/ia-produtividade-lidera-valor-10-liberam-agentes-para-decidir-em-varias-areas/</link><guid isPermaLink="true">https://guics.me/en/artigos/ia-produtividade-lidera-valor-10-liberam-agentes-para-decidir-em-varias-areas/</guid><description>In a global survey released on November 4, 84% of the executives consulted by TCS said productivity is the biggest gain they see in AI. Only 10%…</description><pubDate>Wed, 04 Nov 2026 15:00:00 GMT</pubDate><author>Guilherme da Cruz Silveira</author></item><item><title>HCLTech: AI agents handled 40% more IT incidents</title><link>https://guics.me/en/artigos/hcltech-relata-que-agentes-de-ia-ajudaram-distribuidor-a-absorver-40-mais-incidentes-de-ti/</link><guid isPermaLink="true">https://guics.me/en/artigos/hcltech-relata-que-agentes-de-ia-ajudaram-distribuidor-a-absorver-40-mais-incidentes-de-ti/</guid><description>In a case study published on November 2, HCLTech describes two agents integrated into the support system of a global electronics distributor. The…</description><pubDate>Tue, 03 Nov 2026 15:00:00 GMT</pubDate><author>Guilherme da Cruz Silveira</author></item><item><title>When AI checks the paperwork, the queue moves</title><link>https://guics.me/en/artigos/quando-a-ia-confere-a-papelada-a-fila-anda/</link><guid isPermaLink="true">https://guics.me/en/artigos/quando-a-ia-confere-a-papelada-a-fila-anda/</guid><description>Three English councils tested a Google DeepMind AI that reviews applications for permission to renovate homes: it checks the documents, cross-references…</description><pubDate>Tue, 03 Nov 2026 15:00:00 GMT</pubDate><author>Guilherme da Cruz Silveira</author></item><item><title>Cognizant aligns more than 25,000 professionals with OpenAI AI projects</title><link>https://guics.me/en/artigos/cognizant-alinha-mais-de-25-mil-profissionais-a-projetos-de-ia-com-openai/</link><guid isPermaLink="true">https://guics.me/en/artigos/cognizant-alinha-mais-de-25-mil-profissionais-a-projetos-de-ia-com-openai/</guid><description>Completing learning tracks records participation; the effect needs to be tracked in the operation. In one case, Cognizant credits an assistant with a 60% reduction in…</description><pubDate>Fri, 30 Oct 2026 15:00:00 GMT</pubDate><author>Guilherme da Cruz Silveira</author></item><item><title>Telstra uses AI to reconcile billing during a migration between systems</title><link>https://guics.me/en/artigos/telstra-usa-ia-para-conciliar-cobrancas-durante-migracao-entre-sistemas/</link><guid isPermaLink="true">https://guics.me/en/artigos/telstra-usa-ia-para-conciliar-cobrancas-durante-migracao-entre-sistemas/</guid><description>During platform migrations, charges and payments can end up recorded in different systems. Telstra uses AI to flag discrepancies and…</description><pubDate>Thu, 29 Oct 2026 15:00:00 GMT</pubDate><author>Guilherme da Cruz Silveira</author></item><item><title>Itaú signs deal to buy NeoSpace</title><link>https://guics.me/en/artigos/itau-assina-acordo-para-comprar-a-neospace-apos-quase-dois-anos-de-parceria-em-ia/</link><guid isPermaLink="true">https://guics.me/en/artigos/itau-assina-acordo-para-comprar-a-neospace-apos-quase-dois-anos-de-parceria-em-ia/</guid><description>Itaú signed an agreement to acquire the Brazilian AI company NeoSpace for R$ 2.925 billion. The startup was already developing models with the bank&#x27;s data for areas such as credit, customer relationships and fraud prevention. The deal depends on regulatory approvals; the results Itaú cites don&#x27;t come with public indicators.</description><pubDate>Wed, 28 Oct 2026 15:00:00 GMT</pubDate><author>Guilherme Silveira</author></item><item><title>Who decides when AI gives its opinion?</title><link>https://guics.me/en/artigos/quem-decide-quando-a-ia-da-a-opiniao/</link><guid isPermaLink="true">https://guics.me/en/artigos/quem-decide-quando-a-ia-da-a-opiniao/</guid><description>AI helps you decide better, as long as the decision stays yours. In an Anthropic study with 50 experienced analysts, each one worked with an AI agent that argued, assigned at random, for one of two options in four choices in the analysis. They followed the agent in 86% of the choices.</description><pubDate>Wed, 28 Oct 2026 15:00:00 GMT</pubDate><author>Guilherme da Cruz Silveira</author></item><item><title>ARIS launches a platform to give AI agents process context</title><link>https://guics.me/en/artigos/aris-lanca-plataforma-para-dar-contexto-de-processos-a-agentes-de-ia/</link><guid isPermaLink="true">https://guics.me/en/artigos/aris-lanca-plataforma-para-dar-contexto-de-processos-a-agentes-de-ia/</guid><description>Available since 10/26, the platform combines process mining, opportunity analysis and integration through the Model Context Protocol (MCP), letting agents look up the company&#x27;s rules and systems. For responsible adoption, it&#x27;s worth checking whether records reflect the routine and setting up a measurable pilot. The announcement brings no results from the platform&#x27;s use at clients.</description><pubDate>Tue, 27 Oct 2026 15:00:00 GMT</pubDate><author>Guilherme Silveira</author></item><item><title>Is your store ready to sell to an AI agent?</title><link>https://guics.me/en/artigos/sua-loja-esta-pronta-para-vender-a-um-agente-de-ia/</link><guid isPermaLink="true">https://guics.me/en/artigos/sua-loja-esta-pronta-para-vender-a-um-agente-de-ia/</guid><description>Your store&#x27;s next sale may be made by a robot, and the site needs to let it buy. On the network of Forter, a fraud prevention company, orders placed by AI agents grew about 5 times from late July to October, almost all of them in North America.</description><pubDate>Tue, 27 Oct 2026 15:00:00 GMT</pubDate><author>Guilherme da Cruz Silveira</author></item><item><title>Sicoob Crediauc tests an AI agent in Desenrola debt renegotiations</title><link>https://guics.me/en/artigos/sicoob-crediauc-testa-agente-de-ia-em-renegociacoes-do-desenrola/</link><guid isPermaLink="true">https://guics.me/en/artigos/sicoob-crediauc-testa-agente-de-ia-em-renegociacoes-do-desenrola/</guid><description>In five days, an AI agent started contacts and an employee negotiated and recorded six agreements, according to Ubots. The vendor claims 15 times higher productivity, but the data doesn&#x27;t allow reproducing the math or attributing that difference to the tool. The amount renegotiated per day in the test was about 7.6 times that of the operation it was compared with.</description><pubDate>Mon, 26 Oct 2026 15:00:00 GMT</pubDate><author>Guilherme Silveira</author></item><item><title>How to avoid the technician&#x27;s wasted trip?</title><link>https://guics.me/en/artigos/como-evitar-a-viagem-perdida-do-tecnico-case-algar/</link><guid isPermaLink="true">https://guics.me/en/artigos/como-evitar-a-viagem-perdida-do-tecnico-case-algar/</guid><description>Algar started confirming technical visits on WhatsApp before sending the crew out. Before, it notified only a sample of visits, and the response was 5% of the total. Now it notifies all of them, tries up to three contacts, and the response has reached 68%. Wasted visits fell 30%.</description><pubDate>Mon, 26 Oct 2026 15:00:00 GMT</pubDate><author>Guilherme da Cruz Silveira</author></item><item><title>Companies get more out of AI when they redesign the work, McKinsey shows</title><link>https://guics.me/en/artigos/empresa-rende-mais-quando-redesenha-o-trabalho-com-ia-mckinsey/</link><guid isPermaLink="true">https://guics.me/en/artigos/empresa-rende-mais-quando-redesenha-o-trabalho-com-ia-mckinsey/</guid><description>McKinsey split companies into three groups: those that only give AI tools to employees, those that automate processes and those that redesign work around AI. In the last group, 48% reported a significant gain. In the first two, 13% and 24%, according to the consultancy.</description><pubDate>Sun, 25 Oct 2026 15:00:00 GMT</pubDate><author>Guilherme da Cruz Silveira</author></item><item><title>Nestlé resolves 40% of customer questions without an agent</title><link>https://guics.me/en/artigos/nestle-resolve-40-das-duvidas-de-clientes-sem-atendente/</link><guid isPermaLink="true">https://guics.me/en/artigos/nestle-resolve-40-das-duvidas-de-clientes-sem-atendente/</guid><description>Anyone who has fought with a customer service bot has reason to distrust AI in this service. Today&#x27;s platforms deliver a different result. At Nestlé, where more than half of contacts come in through chat, email, WhatsApp and forms, the pilot resolved 40% of cases without a human agent, with a rating of 4.8 out of 5.</description><pubDate>Sun, 25 Oct 2026 15:00:00 GMT</pubDate><author>Guilherme da Cruz Silveira</author></item><item><title>A stalled quote cools the sale; AI was the way out</title><link>https://guics.me/en/artigos/orcamento-parado-esfria-a-venda-mitsubishi-electric-ia/</link><guid isPermaLink="true">https://guics.me/en/artigos/orcamento-parado-esfria-a-venda-mitsubishi-electric-ia/</guid><description>Mitsubishi Electric Digital Innovation, part of the Mitsubishi Electric group in Japan, receives about 100 quote requests a month. The customer was left waiting, and the technician redid the math when information was missing. In two months, the company built an AI agent that prepares the draft of each quote.</description><pubDate>Sat, 24 Oct 2026 15:00:00 GMT</pubDate><author>Guilherme da Cruz Silveira</author></item><item><title>How a giant&#x27;s finance team cut 80% of the time on a repetitive task</title><link>https://guics.me/en/artigos/como-o-financeiro-de-uma-gigante-cortou-80-do-tempo/</link><guid isPermaLink="true">https://guics.me/en/artigos/como-o-financeiro-de-uma-gigante-cortou-80-do-tempo/</guid><description>Centalion (formerly Gunvor), a Geneva commodities trader that had revenue of US$ 144 billion in 2025, put an AI agent in its treasury.</description><pubDate>Fri, 23 Oct 2026 15:00:00 GMT</pubDate><author>Guilherme da Cruz Silveira</author></item><item><title>Uber expands ChatGPT Work to over 30,000 employees</title><link>https://guics.me/en/artigos/uber-amplia-chatgpt-work-para-mais-de-30-mil-funcionarios-retorno-segue-em-avaliacao/</link><guid isPermaLink="true">https://guics.me/en/artigos/uber-amplia-chatgpt-work-para-mais-de-30-mil-funcionarios-retorno-segue-em-avaliacao/</guid><description>Uber announced access to ChatGPT Work for more than 30,000 employees after a pilot with 1,000 people. The company has not yet released measured gains or the…</description><pubDate>Fri, 23 Oct 2026 15:00:00 GMT</pubDate><author>Guilherme da Cruz Silveira</author></item><item><title>Security alerts: 3 questions to ask your IT team</title><link>https://guics.me/en/artigos/alertas-de-seguranca-3-perguntas-para-fazer-a-equipe-de-ti/</link><guid isPermaLink="true">https://guics.me/en/artigos/alertas-de-seguranca-3-perguntas-para-fazer-a-equipe-de-ti/</guid><description>On October 21 and 22, 2026, technology companies released alerts about websites, apps, AI tools and access credentials stored in the cloud. That doesn&#x27;t mean your company was affected. Here are three questions to pass on to whoever takes care of your systems.</description><pubDate>Thu, 22 Oct 2026 15:00:00 GMT</pubDate><author>Guilherme Silveira</author></item><item><title>Your customer will buy by chatting with an AI. Is your store ready?</title><link>https://guics.me/en/artigos/seu-cliente-vai-comprar-conversando-com-uma-ia/</link><guid isPermaLink="true">https://guics.me/en/artigos/seu-cliente-vai-comprar-conversando-com-uma-ia/</guid><description>Meta said on 10/21, when it released its quarterly results, that Muse, its AI agent, passed 10 million users and that shopping is among its main uses. Buying through conversation is already a habit, and the company that answers fast and with the right information gets ahead.</description><pubDate>Thu, 22 Oct 2026 15:00:00 GMT</pubDate><author>Guilherme da Cruz Silveira</author></item><item><title>Google brings brand, finance and Gemini apps into a hub for small businesses</title><link>https://guics.me/en/artigos/google-reune-marca-financas-e-apps-do-gemini-em-hub-para-pequenos-negocios/</link><guid isPermaLink="true">https://guics.me/en/artigos/google-reune-marca-financas-e-apps-do-gemini-em-hub-para-pequenos-negocios/</guid><description>Google announced a Gemini space that brings together brand tasks, financial analysis and connections to apps. For a small business, the practical question is whether a specific routine gets faster without increasing the rework of checking.</description><pubDate>Wed, 21 Oct 2026 15:00:00 GMT</pubDate><author>Guilherme Silveira</author></item><item><title>Egnyte organizes context and access for AI agents</title><link>https://guics.me/en/artigos/egnyte-lanca-ferramentas-para-organizar-o-contexto-e-controlar-o-acesso-de-agentes-de-ia/</link><guid isPermaLink="true">https://guics.me/en/artigos/egnyte-lanca-ferramentas-para-organizar-o-contexto-e-controlar-o-acesso-de-agentes-de-ia/</guid><description>Egnyte launched features to show which information feeds AI agents and to limit what they access. See what this teaches about context.</description><pubDate>Mon, 19 Oct 2026 15:00:00 GMT</pubDate><author>Guilherme da Cruz Silveira</author></item><item><title>Parallel agents in ChatGPT: how to organize a first test at your company</title><link>https://guics.me/en/artigos/agentes-em-paralelo-no-chatgpt-como-organizar-um-primeiro-teste-na-empresa/</link><guid isPermaLink="true">https://guics.me/en/artigos/agentes-em-paralelo-no-chatgpt-como-organizar-um-primeiro-teste-na-empresa/</guid><description>The GPT-6 Pro update lets you split tasks between agents. Here&#x27;s how to organize an internal test with authorized sources, verifiable deliverables and human review.</description><pubDate>Sat, 17 Oct 2026 15:00:00 GMT</pubDate><author>Guilherme Silveira</author></item><item><title>Cohere signs letter of intent to acquire security startup AISLE</title><link>https://guics.me/en/artigos/cohere-assina-carta-de-intencao-para-adquirir-startup-aisle-de-seguranca/</link><guid isPermaLink="true">https://guics.me/en/artigos/cohere-assina-carta-de-intencao-para-adquirir-startup-aisle-de-seguranca/</guid><description>Cohere signed a letter of intent to acquire AISLE, a security startup that uses AI to find vulnerabilities. Reuters reporting, republished on Investing.com, credits AISLE with discovering more than 500 new flaws in widely used software in its first year, including 12 in OpenSSL, fixed after discovery. The transaction has not closed yet.</description><pubDate>Thu, 15 Oct 2026 15:00:00 GMT</pubDate><author>Guilherme Silveira</author></item><item><title>SolarWinds opens guided access to Avior, an AI agent for IT operations</title><link>https://guics.me/en/artigos/solarwinds-abre-acesso-guiado-ao-avior-agente-de-ia-para-operacoes-de-ti/</link><guid isPermaLink="true">https://guics.me/en/artigos/solarwinds-abre-acesso-guiado-ao-avior-agente-de-ia-para-operacoes-de-ti/</guid><description>Announced on October 14, 2026, Avior brings together agents aimed at IT operations goals, such as investigating alerts and finding configuration drift. The network engineering agent is available through guided onboarding. Examples and controls are the vendor&#x27;s statements; no independent performance evaluation has been published.</description><pubDate>Thu, 15 Oct 2026 15:00:00 GMT</pubDate><author>Guilherme Silveira</author></item><item><title>Seismic expands sales agents and takes AI workflows to Slack, Copilot, Claude</title><link>https://guics.me/en/artigos/seismic-amplia-agentes-comerciais-e-leva-fluxos-de-ia-a-slack-copilot-e-claude/</link><guid isPermaLink="true">https://guics.me/en/artigos/seismic-amplia-agentes-comerciais-e-leva-fluxos-de-ia-a-slack-copilot-e-claude/</guid><description>At the Seismic Shift conference, held on October 13, 2026, the company announced two agents for sales teams and new integrations with Slack, Microsoft 365 Copilot and Claude. Deal Agent is already available in Seismic and in Highspot by Seismic. The announcement describes product features, without presenting independent conversion or revenue results.</description><pubDate>Thu, 15 Oct 2026 15:00:00 GMT</pubDate><author>Guilherme Silveira</author></item><item><title>Gemini 4 Argon: your company&#x27;s test has to include the December price</title><link>https://guics.me/en/artigos/gemini-4-argon-o-teste-da-empresa-precisa-incluir-o-preco-de-dezembro/</link><guid isPermaLink="true">https://guics.me/en/artigos/gemini-4-argon-o-teste-da-empresa-precisa-incluir-o-preco-de-dezembro/</guid><description>Google announced an introductory price for Gemini 4 Argon until November 30. Here&#x27;s how to compare the two cost scenarios and evaluate whether to keep using it at your company.</description><pubDate>Thu, 15 Oct 2026 15:00:00 GMT</pubDate><author>Guilherme Silveira</author></item><item><title>Microsoft-Decision-1: how to use one AI to evaluate another&#x27;s answers</title><link>https://guics.me/en/artigos/microsoft-decision-1-como-usar-uma-ia-para-avaliar-respostas-de-outra/</link><guid isPermaLink="true">https://guics.me/en/artigos/microsoft-decision-1-como-usar-uma-ia-para-avaliar-respostas-de-outra/</guid><description>Microsoft-Decision-1 lets you evaluate AI answers according to the criteria and evidence provided. Here&#x27;s how to scope its use, test the evaluator and keep human review.</description><pubDate>Tue, 13 Oct 2026 15:00:00 GMT</pubDate><author>Guilherme Silveira</author></item><item><title>Anthropic details AI usage rules; new version takes effect in November</title><link>https://guics.me/en/artigos/anthropic-detalha-regras-de-uso-da-ia-nova-versao-vale-em-novembro/</link><guid isPermaLink="true">https://guics.me/en/artigos/anthropic-detalha-regras-de-uso-da-ia-nova-versao-vale-em-novembro/</guid><description>The update brings together examples for deceptive campaigns and high-risk uses. Among the controls spelled out are qualified human review and notice to affected people, requirements that, according to the company, already existed. The new version of Claude&#x27;s policy takes effect on November 12; it is not a new Brazilian law.</description><pubDate>Tue, 13 Oct 2026 15:00:00 GMT</pubDate><author>Guilherme Silveira</author></item><item><title>Maia 300: a promise of more tokens per megawatt doesn&#x27;t set the cost of AI</title><link>https://guics.me/en/artigos/maia-300-promessa-de-mais-tokens-por-megawatt-nao-define-o-custo-da-ia/</link><guid isPermaLink="true">https://guics.me/en/artigos/maia-300-promessa-de-mais-tokens-por-megawatt-nao-define-o-custo-da-ia/</guid><description>The announcement includes the vendor&#x27;s own projections for datacenter efficiency, but doesn&#x27;t say what a customer will pay for an adequate answer. The buying decision depends on comparing cost per accepted task, quality and availability, under conditions similar to real use.</description><pubDate>Tue, 13 Oct 2026 15:00:00 GMT</pubDate><author>Guilherme Silveira</author></item><item><title>ICO opens consultation on data risks when AI agents carry out tasks</title><link>https://guics.me/en/artigos/ico-poe-em-consulta-os-riscos-de-dados-quando-agentes-de-ia-executam-tarefas/</link><guid isPermaLink="true">https://guics.me/en/artigos/ico-poe-em-consulta-os-riscos-de-dados-quando-agentes-de-ia-executam-tarefas/</guid><description>When an agent starts consulting tools and systems, data control continues after deployment. The British regulator opened a consultation on these risks and reported changes agreed with ten developers. The process will guide future guidance in the United Kingdom; the consultation, by itself, doesn&#x27;t create a new obligation for Brazilian companies.</description><pubDate>Fri, 09 Oct 2026 15:00:00 GMT</pubDate><author>Guilherme Silveira</author></item><item><title>OpenAI identified almost 100 articles under fake journalist identities</title><link>https://guics.me/en/artigos/openai-diz-ter-identificado-quase-100-artigos-sob-identidades-jornalisticas-falsas/</link><guid isPermaLink="true">https://guics.me/en/artigos/openai-diz-ter-identificado-quase-100-artigos-sob-identidades-jornalisticas-falsas/</guid><description>An OpenAI report describes two influence operations that combined AI models and traditional tactics. In the Iranian operation, the company says it found almost a hundred articles published or distributed under seven journalist identities. The account distinguishes circulation in outlets from traction on social media and records the limits of verification.</description><pubDate>Fri, 09 Oct 2026 15:00:00 GMT</pubDate><author>Guilherme Silveira</author></item><item><title>Claude in Google Workspace: how to organize an AI editing pilot</title><link>https://guics.me/en/artigos/claude-no-google-workspace-como-organizar-um-piloto-de-edicao-com-ia/</link><guid isPermaLink="true">https://guics.me/en/artigos/claude-no-google-workspace-como-organizar-um-piloto-de-edicao-com-ia/</guid><description>Here&#x27;s how to structure an editing pilot with Claude in Google Workspace, defining owners, limits, metrics and version recovery.</description><pubDate>Thu, 08 Oct 2026 15:00:00 GMT</pubDate><author>Guilherme Silveira</author></item><item><title>IBM and Red Hat fixed more than 400 vulnerabilities in Java libraries</title><link>https://guics.me/en/artigos/ibm-e-red-hat-dizem-ter-corrigido-mais-de-400-vulnerabilidades-em-bibliotecas-java/</link><guid isPermaLink="true">https://guics.me/en/artigos/ibm-e-red-hat-dizem-ter-corrigido-mais-de-400-vulnerabilidades-em-bibliotecas-java/</guid><description>IBM and Red Hat say the Lightwell initiative fixed more than 400 vulnerabilities in Java libraries. The announcement doesn&#x27;t say which components or versions were affected, so it doesn&#x27;t allow concluding that a specific system is exposed. For companies that depend on third-party software, the answer starts with the vendor.</description><pubDate>Thu, 08 Oct 2026 15:00:00 GMT</pubDate><author>Guilherme Silveira</author></item><item><title>ChatGPT starts building calculators and forms inside the conversation</title><link>https://guics.me/en/artigos/chatgpt-comeca-a-montar-calculadoras-e-formularios-dentro-da-conversa/</link><guid isPermaLink="true">https://guics.me/en/artigos/chatgpt-comeca-a-montar-calculadoras-e-formularios-dentro-da-conversa/</guid><description>GPT-6 started generating charts, forms and interactive tools in ChatGPT. The feature lets teams explore comparisons without leaving the conversation. The rollout starts on paid plans; OpenAI has not published productivity gain data.</description><pubDate>Thu, 08 Oct 2026 15:00:00 GMT</pubDate><author>Guilherme Silveira</author></item><item><title>Anthropic announces training with a real project to prepare AI engineers</title><link>https://guics.me/en/artigos/anthropic-anuncia-formacao-com-projeto-real-para-preparar-engenheiros-de-ia/</link><guid isPermaLink="true">https://guics.me/en/artigos/anthropic-anuncia-formacao-com-projeto-real-para-preparar-engenheiros-de-ia/</guid><description>Anthropic announced a program that combines technical preparation, practical assessment and work on a project at the participating organization. The initiative aims to train professionals capable of deploying AI inside companies. Its scale is a target through 2027; the announcement doesn&#x27;t yet show productivity results or return on the projects.</description><pubDate>Tue, 06 Oct 2026 15:00:00 GMT</pubDate><author>Guilherme Silveira</author></item><item><title>AI agent test checks the records left in the system after the task</title><link>https://guics.me/en/artigos/teste-de-agentes-de-ia-confere-os-registros-deixados-no-sistema-apos-a-tarefa/</link><guid isPermaLink="true">https://guics.me/en/artigos/teste-de-agentes-de-ia-confere-os-registros-deixados-no-sistema-apos-a-tarefa/</guid><description>Microsoft and Hugging Face released an evaluation that checks the changes agents leave in simulated systems. The study brings testing closer to the operational consequence of each action and examines whether it repeats. The results belong to the environment evaluated; for companies, they offer a criterion for checking tasks before expanding autonomy.</description><pubDate>Tue, 06 Oct 2026 15:00:00 GMT</pubDate><author>Guilherme Silveira</author></item><item><title>Together launches a tool to switch the AI model in coding</title><link>https://guics.me/en/artigos/together-lanca-ferramenta-para-trocar-o-modelo-de-ia-usado-no-desenvolvimento-de-software/</link><guid isPermaLink="true">https://guics.me/en/artigos/together-lanca-ferramenta-para-trocar-o-modelo-de-ia-usado-no-desenvolvimento-de-software/</guid><description>Together AI presented a tool that lets you use other models in coding agents the team already knows. The move opens room to compare vendors with less change to the routine. The savings the company announces depend on your own evaluation, including quality, review and the cost of the finished deliverable.</description><pubDate>Tue, 06 Oct 2026 15:00:00 GMT</pubDate><author>Guilherme Silveira</author></item><item><title>ChatGPT Space at your company: access, review and accountability</title><link>https://guics.me/en/artigos/chatgpt-space-na-empresa-acesso-revisao-e-responsabilidade/</link><guid isPermaLink="true">https://guics.me/en/artigos/chatgpt-space-na-empresa-acesso-revisao-e-responsabilidade/</guid><description>How to organize access, sources, review and owners in ChatGPT Space to support the team&#x27;s decisions and check recurring updates.</description><pubDate>Mon, 05 Oct 2026 15:00:00 GMT</pubDate><author>Guilherme Silveira</author></item><item><title>Decision models: how to test AI before automating a process</title><link>https://guics.me/en/artigos/modelos-de-decisao-como-testar-a-ia-antes-de-automatizar-um-processo/</link><guid isPermaLink="true">https://guics.me/en/artigos/modelos-de-decisao-como-testar-a-ia-antes-de-automatizar-um-processo/</guid><description>A roadmap for testing decision models like Clef: scoping steps, weighing the cost of errors and defining exceptions before automating processes.</description><pubDate>Fri, 02 Oct 2026 15:00:00 GMT</pubDate><author>Guilherme Silveira</author></item><item><title>OpenAI&#x27;s dots: what leaders need to define before delegating to AI</title><link>https://guics.me/en/artigos/dots-da-openai-o-que-lideres-precisam-definir-antes-de-delegar-a-ia/</link><guid isPermaLink="true">https://guics.me/en/artigos/dots-da-openai-o-que-lideres-precisam-definir-antes-de-delegar-a-ia/</guid><description>Here&#x27;s how to evaluate OpenAI&#x27;s dots in the operation: choosing the process, data access, owners, approval and indicators for a pilot.</description><pubDate>Thu, 01 Oct 2026 15:00:00 GMT</pubDate><author>Guilherme Silveira</author></item><item><title>Would you let an agent negotiate for you?</title><link>https://guics.me/en/artigos/voce-deixaria-um-agente-negociar-por-voce-o-experimento-que-expoe-o-limite-da-ia/</link><guid isPermaLink="true">https://guics.me/en/artigos/voce-deixaria-um-agente-negociar-por-voce-o-experimento-que-expoe-o-limite-da-ia/</guid><description>Anthropic had agents trade books on behalf of 201 employees. The biggest problem wasn&#x27;t negotiating. See what the experiment teaches companies.</description><pubDate>Fri, 25 Sep 2026 15:00:00 GMT</pubDate><author>Guilherme da Cruz Silveira</author></item><item><title>AI governance starts with claims that can be tested</title><link>https://guics.me/en/artigos/governanca-de-ia-comeca-por-claims-que-podem-ser-testados/</link><guid isPermaLink="true">https://guics.me/en/artigos/governanca-de-ia-comeca-por-claims-que-podem-ser-testados/</guid><description>OpenAI published principles for independent AI assessments: scope, access, method, independence and correction. See what to take to your company.</description><pubDate>Wed, 23 Sep 2026 15:00:00 GMT</pubDate><author>Guilherme da Cruz Silveira</author></item><item><title>AI is also an infrastructure problem</title><link>https://guics.me/en/artigos/ia-tambem-e-um-problema-de-infraestrutura/</link><guid isPermaLink="true">https://guics.me/en/artigos/ia-tambem-e-um-problema-de-infraestrutura/</guid><description>EPE linked AI, data centers, chips, connectivity and energy in a TechGrid 2026 panel. See why this affects cost and vendor choice.</description><pubDate>Tue, 22 Sep 2026 15:00:00 GMT</pubDate><author>Guilherme da Cruz Silveira</author></item><item><title>Continuous security enters the development cycle with agents</title><link>https://guics.me/en/artigos/seguranca-continua-entra-no-ciclo-de-desenvolvimento-com-agentes/</link><guid isPermaLink="true">https://guics.me/en/artigos/seguranca-continua-entra-no-ciclo-de-desenvolvimento-com-agentes/</guid><description>Google described agents that examine code changes, validate vulnerabilities and propose fixes before production. See what to adapt.</description><pubDate>Tue, 22 Sep 2026 15:00:00 GMT</pubDate><author>Guilherme da Cruz Silveira</author></item><item><title>Sony Bank cuts its development cycle by 30% by applying AI to its core system</title><link>https://guics.me/en/artigos/sony-bank-reduz-30-do-ciclo-de-desenvolvimento-ao-aplicar-ia-no-sistema-central/</link><guid isPermaLink="true">https://guics.me/en/artigos/sony-bank-reduz-30-do-ciclo-de-desenvolvimento-ao-aplicar-ia-no-sistema-central/</guid><description>Sony Bank and Fujitsu report 30% less time between design and testing by applying AI to a banking system in operation. See the conditions behind the result.</description><pubDate>Mon, 21 Sep 2026 15:00:00 GMT</pubDate><author>Guilherme da Cruz Silveira</author></item><item><title>The next fight in enterprise AI is over the layer that governs agents</title><link>https://guics.me/en/artigos/a-proxima-disputa-da-ia-empresarial-esta-na-camada-que-governa-os-agentes/</link><guid isPermaLink="true">https://guics.me/en/artigos/a-proxima-disputa-da-ia-empresarial-esta-na-camada-que-governa-os-agentes/</guid><description>The enterprise AI fight moves from models to the layer that lets agents understand the business, access systems and act with rules and control.</description><pubDate>Fri, 18 Sep 2026 15:00:00 GMT</pubDate><author>Guilherme da Cruz Silveira</author></item><item><title>Recovering an agent&#x27;s state becomes part of AI governance</title><link>https://guics.me/en/artigos/recuperar-o-estado-de-um-agente-vira-parte-da-governanca-de-ia/</link><guid isPermaLink="true">https://guics.me/en/artigos/recuperar-o-estado-de-um-agente-vira-parte-da-governanca-de-ia/</guid><description>Cohesity presented a feature to protect and recover the memory, configuration and resources of AI agents. See why this belongs in the continuity plan.</description><pubDate>Thu, 17 Sep 2026 15:00:00 GMT</pubDate><author>Guilherme da Cruz Silveira</author></item><item><title>AI agents enter regulatory processes, but they need to be traceable</title><link>https://guics.me/en/artigos/agentes-de-ia-entram-em-processos-regulatorios-mas-precisam-ser-rastreaveis/</link><guid isPermaLink="true">https://guics.me/en/artigos/agentes-de-ia-entram-em-processos-regulatorios-mas-precisam-ser-rastreaveis/</guid><description>Workiva presented agents for regulatory reporting and audits. Automating these tasks requires sources, permissions, evidence and owners.</description><pubDate>Wed, 16 Sep 2026 15:00:00 GMT</pubDate><author>Guilherme da Cruz Silveira</author></item><item><title>AI cost: price per token isn&#x27;t enough to choose the model</title><link>https://guics.me/en/artigos/custo-de-ia-preco-por-token-nao-basta-para-decidir-o-modelo/</link><guid isPermaLink="true">https://guics.me/en/artigos/custo-de-ia-preco-por-token-nao-basta-para-decidir-o-modelo/</guid><description>Price per token doesn&#x27;t show the real cost of an AI model. Compare cost per completed task, quality, response time and the review needed.</description><pubDate>Mon, 14 Sep 2026 15:00:00 GMT</pubDate><author>Guilherme da Cruz Silveira</author></item><item><title>AI security: more autonomous attacks call for an inventory before the crisis</title><link>https://guics.me/en/artigos/seguranca-de-ia-ataques-mais-autonomos-exigem-inventario-antes-da-crise/</link><guid isPermaLink="true">https://guics.me/en/artigos/seguranca-de-ia-ataques-mais-autonomos-exigem-inventario-antes-da-crise/</guid><description>More autonomous attacks call for an inventory of credentials, code, instructions, models and compute quotas before the crisis. See where to start.</description><pubDate>Mon, 14 Sep 2026 15:00:00 GMT</pubDate><author>Guilherme da Cruz Silveira</author></item><item><title>ChatGPT and Codex: choose by the work environment</title><link>https://guics.me/en/artigos/chatgpt-e-codex-escolha-pelo-ambiente-de-trabalho/</link><guid isPermaLink="true">https://guics.me/en/artigos/chatgpt-e-codex-escolha-pelo-ambiente-de-trabalho/</guid><description>ChatGPT works best for conversation, research and content. Codex makes sense when there are files, code and tests. See how to choose by the type of work.</description><pubDate>Fri, 28 Aug 2026 15:00:00 GMT</pubDate><author>Guilherme da Cruz Silveira</author></item><item><title>Human control is still part of the responsibility in using AI</title><link>https://guics.me/en/artigos/controle-humano-continua-sendo-parte-da-responsabilidade-no-uso-de-ia/</link><guid isPermaLink="true">https://guics.me/en/artigos/controle-humano-continua-sendo-parte-da-responsabilidade-no-uso-de-ia/</guid><description>OAB debates reinforce transparency, data protection and human control in the use of AI. Gaining speed doesn&#x27;t transfer responsibility to the system.</description><pubDate>Thu, 27 Aug 2026 15:00:00 GMT</pubDate><author>Guilherme da Cruz Silveira</author></item><item><title>Choosing an AI model also decides who controls access</title><link>https://guics.me/en/artigos/escolher-um-modelo-de-ia-tambem-define-quem-controla-o-acesso/</link><guid isPermaLink="true">https://guics.me/en/artigos/escolher-um-modelo-de-ia-tambem-define-quem-controla-o-acesso/</guid><description>Performance isn&#x27;t enough to choose an AI model. Contract, permissions, data protection and a continuity plan define who controls the use.</description><pubDate>Tue, 18 Aug 2026 15:00:00 GMT</pubDate><author>Guilherme da Cruz Silveira</author></item><item><title>Autonomous agents expand testing; scaling requires limits</title><link>https://guics.me/en/artigos/agentes-autonomos-ampliam-testes-escalar-exige-limites/</link><guid isPermaLink="true">https://guics.me/en/artigos/agentes-autonomos-ampliam-testes-escalar-exige-limites/</guid><description>Autonomous agents speed up experiments, but every long run needs a goal, a budget, permissions and a stopping criterion before it goes into operation.</description><pubDate>Mon, 17 Aug 2026 15:00:00 GMT</pubDate><author>Guilherme da Cruz Silveira</author></item><item><title>Local or external AI: the decision starts with the process</title><link>https://guics.me/en/artigos/ia-local-ou-externa-a-decisao-comeca-no-processo/</link><guid isPermaLink="true">https://guics.me/en/artigos/ia-local-ou-externa-a-decisao-comeca-no-processo/</guid><description>Local AI gives more control over data; an external service brings capacity and flexibility. See how to decide based on the process and the flow of information.</description><pubDate>Mon, 17 Aug 2026 15:00:00 GMT</pubDate><author>Guilherme da Cruz Silveira</author></item><item><title>Before sending data to AI, classify what can leave the company</title><link>https://guics.me/en/artigos/antes-de-enviar-dados-a-ia-classifique-o-que-pode-sair-da-empresa/</link><guid isPermaLink="true">https://guics.me/en/artigos/antes-de-enviar-dados-a-ia-classifique-o-que-pode-sair-da-empresa/</guid><description>A simple way to classify public, internal, personal and sensitive data before using AI, with authorized uses and an owner for each decision.</description><pubDate>Fri, 14 Aug 2026 15:00:00 GMT</pubDate><author>Guilherme da Cruz Silveira</author></item><item><title>AI infrastructure also defines the cost of operating</title><link>https://guics.me/en/artigos/infraestrutura-de-ia-tambem-define-o-custo-de-operar/</link><guid isPermaLink="true">https://guics.me/en/artigos/infraestrutura-de-ia-tambem-define-o-custo-de-operar/</guid><description>Compute, energy, vendors and dependencies all go into the cost of scaling AI. See why the tool&#x27;s initial price doesn&#x27;t tell the whole story.</description><pubDate>Fri, 14 Aug 2026 15:00:00 GMT</pubDate><author>Guilherme da Cruz Silveira</author></item><item><title>Ads in ChatGPT open a discovery channel, but results need testing</title><link>https://guics.me/en/artigos/publicidade-no-chatgpt-abre-canal-de-descoberta-mas-exige-teste-de-resultado/</link><guid isPermaLink="true">https://guics.me/en/artigos/publicidade-no-chatgpt-abre-canal-de-descoberta-mas-exige-teste-de-resultado/</guid><description>An ad in ChatGPT opens a discovery point, but showing up isn&#x27;t a return. See how to test the channel with a defined offer, audience and result measure.</description><pubDate>Fri, 14 Aug 2026 15:00:00 GMT</pubDate><author>Guilherme da Cruz Silveira</author></item></channel></rss>