<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Tech on Althora</title><link>https://althora.com/tags/tech/</link><description>Recent content in Tech on Althora</description><generator>Hugo</generator><language>en-US</language><copyright>© 2026 Althora</copyright><lastBuildDate>Sun, 23 Aug 2026 00:00:00 +0000</lastBuildDate><atom:link href="https://althora.com/tags/tech/index.xml" rel="self" type="application/rss+xml"/><item><title>The Senior DevOps and SRE Job Search in 2026 Has a Math Problem, Not a Talent Problem</title><link>https://althora.com/blog/senior-devops-hiring-2026-math-problem/</link><pubDate>Sun, 23 Aug 2026 00:00:00 +0000</pubDate><guid>https://althora.com/blog/senior-devops-hiring-2026-math-problem/</guid><description>&lt;p&gt;Every senior engineer applying right now already knows the feeling: send fifty applications, hear back from two, and finish the month no closer to an offer than you started. The instinct is to blame the market, or blame yourself. Neither is quite right. What actually happened is more specific: AI made applying free and made screening free, on the same timeline, and nobody made interviewing any cheaper. That asymmetry is the whole story.&lt;/p&gt;</description></item><item><title>Control Tower: the panel I built because fifteen open fronts isn't a discipline problem</title><link>https://althora.com/blog/control-tower-gtd/</link><pubDate>Sat, 22 Aug 2026 00:00:00 +0000</pubDate><guid>https://althora.com/blog/control-tower-gtd/</guid><description>&lt;p&gt;Every time a task falls through, the obvious explanation is a discipline problem. It almost never is. In most cases, what slips away is whatever never had a written next action, or never had a date attached to chase it.&lt;/p&gt;
&lt;p&gt;Running several fronts at once — business, product, infrastructure, plus whatever personal stuff can&amp;rsquo;t wait — I got to a point where the list of open projects passed ten and no single place showed all of them at the same time. The feeling of drowning didn&amp;rsquo;t come from volume. It came from not having one place to look.&lt;/p&gt;</description></item><item><title>O que rodar IA dentro de um escritório de advocacia tem em comum com qualquer sistema que não pode "ligar pra casa"</title><link>https://althora.com/pt/blog/ia-on-prem-nao-liga-pra-casa/</link><pubDate>Sat, 22 Aug 2026 00:00:00 +0000</pubDate><guid>https://althora.com/pt/blog/ia-on-prem-nao-liga-pra-casa/</guid><description>&lt;p&gt;Quem já operou um sistema que não pode sair da rede — um ambiente air-gapped, um appliance num cliente sem regra de saída no firewall, um deploy regulado que não pode &amp;ldquo;ligar pra casa&amp;rdquo; — reconhece um padrão de longe. Não é um padrão de IA. É rotina antiga de infraestrutura, e a Anne for Legal é só a versão mais recente dele.&lt;/p&gt;
&lt;h2 id="o-problema-não-é-novo-só-o-componente-que-falha-é"&gt;O problema não é novo, só o componente que falha é&lt;/h2&gt;
&lt;p&gt;Todo sistema que precisa continuar funcionando sem depender de algo externo tem que responder a uma pergunta desconfortável: &lt;strong&gt;o que acontece quando a peça que você mais queria que funcionasse, não funciona?&lt;/strong&gt;&lt;/p&gt;</description></item><item><title>Torre de Controle: o painel que montei porque 15 frentes abertas não é um problema de disciplina</title><link>https://althora.com/pt/blog/torre-de-controle-gtd/</link><pubDate>Sat, 22 Aug 2026 00:00:00 +0000</pubDate><guid>https://althora.com/pt/blog/torre-de-controle-gtd/</guid><description>&lt;p&gt;Toda vez que uma tarefa some, a explicação óbvia é falta de disciplina. Quase nunca é essa. Na maioria dos casos, o que escapa é o que nunca teve uma próxima ação escrita, ou nunca teve uma data marcada para ser cobrado.&lt;/p&gt;
&lt;p&gt;Eu tocando várias frentes ao mesmo tempo — comercial, produto, infraestrutura, mais o que é pessoal e não pode esperar — cheguei num ponto em que a lista de projetos abertos passava de dez e nenhum lugar mostrava todos ao mesmo tempo. A sensação de estar &amp;ldquo;afogado&amp;rdquo; não vinha do volume. Vinha de não ter um lugar único pra olhar.&lt;/p&gt;</description></item><item><title>What running AI inside a law firm has in common with anything that can't "phone home"</title><link>https://althora.com/blog/on-prem-ai-appliance-devops/</link><pubDate>Sat, 22 Aug 2026 00:00:00 +0000</pubDate><guid>https://althora.com/blog/on-prem-ai-appliance-devops/</guid><description>&lt;p&gt;Anyone who has operated a system that can&amp;rsquo;t leave the network — an air-gapped environment, an appliance at a client site with no outbound firewall rule, a regulated deployment that isn&amp;rsquo;t allowed to &amp;ldquo;phone home&amp;rdquo; — recognizes this pattern from a mile away. It isn&amp;rsquo;t an AI pattern. It&amp;rsquo;s old infrastructure routine, and Anne for Legal is just its latest instance.&lt;/p&gt;
&lt;h2 id="the-problem-isnt-new-only-the-failing-component-is"&gt;The problem isn&amp;rsquo;t new, only the failing component is&lt;/h2&gt;
&lt;p&gt;Any system that needs to keep working without depending on something external has to answer an uncomfortable question: &lt;strong&gt;what happens when the one piece you most wanted to work, doesn&amp;rsquo;t?&lt;/strong&gt;&lt;/p&gt;</description></item><item><title>Grok Bot vs. Claude Cowork: onde as coisas realmente estão</title><link>https://althora.com/pt/blog/grok-bot-vs-claude-cowork/</link><pubDate>Fri, 21 Aug 2026 00:00:00 +0000</pubDate><guid>https://althora.com/pt/blog/grok-bot-vs-claude-cowork/</guid><description>&lt;p&gt;&amp;ldquo;Grok vs Claude&amp;rdquo; costuma ser tratado como uma única comparação. Na verdade são duas, e cada uma tem uma resposta diferente.&lt;/p&gt;
&lt;ol&gt;
&lt;li&gt;&lt;strong&gt;Modelo vs. modelo&lt;/strong&gt; — o Grok 4.6 da xAI contra o Claude Opus 5 / Sonnet 5 da Anthropic. É a comparação de API e chatbot.&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;Produto de agente vs. produto de agente&lt;/strong&gt; — &lt;strong&gt;Grok Bot&lt;/strong&gt; contra &lt;strong&gt;Claude Cowork&lt;/strong&gt;. São os produtos &amp;ldquo;dá o trabalho e sai de perto&amp;rdquo;, e a arquitetura dos dois é bem diferente.&lt;/li&gt;
&lt;/ol&gt;
&lt;p&gt;Quando alguém diz &amp;ldquo;Grok Bot&amp;rdquo;, está falando especificamente do item 2 — não do chatbot.&lt;/p&gt;</description></item><item><title>Grok Bot vs. Claude Cowork: Where Things Actually Stand</title><link>https://althora.com/blog/grok-bot-vs-claude-cowork/</link><pubDate>Fri, 21 Aug 2026 00:00:00 +0000</pubDate><guid>https://althora.com/blog/grok-bot-vs-claude-cowork/</guid><description>&lt;p&gt;&amp;ldquo;Grok vs Claude&amp;rdquo; gets asked as if it were one comparison. It&amp;rsquo;s really two, and they have different answers.&lt;/p&gt;
&lt;ol&gt;
&lt;li&gt;&lt;strong&gt;Model vs. model&lt;/strong&gt; — xAI&amp;rsquo;s Grok 4.6 against Anthropic&amp;rsquo;s Claude Opus 5 / Sonnet 5. This is the API-and-chatbot comparison.&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;Agent product vs. agent product&lt;/strong&gt; — &lt;strong&gt;Grok Bot&lt;/strong&gt; against &lt;strong&gt;Claude Cowork&lt;/strong&gt;. These are the &amp;ldquo;give it a job and walk away&amp;rdquo; products, and they are architecturally quite different.&lt;/li&gt;
&lt;/ol&gt;
&lt;p&gt;If someone says &amp;ldquo;Grok Bot,&amp;rdquo; they mean #2 specifically — not the chatbot.&lt;/p&gt;</description></item></channel></rss>