<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>AI on Althora</title><link>https://althora.com/tags/ai/</link><description>Recent content in AI on Althora</description><generator>Hugo</generator><language>en-US</language><copyright>© 2026 Althora</copyright><lastBuildDate>Sat, 22 Aug 2026 00:00:00 +0000</lastBuildDate><atom:link href="https://althora.com/tags/ai/index.xml" rel="self" type="application/rss+xml"/><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>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><item><title>Fake AI Tools Are Now a Top Malware Lure for Small Businesses</title><link>https://althora.com/blog/fake-ai-tools-malware-small-business/</link><pubDate>Fri, 07 Aug 2026 00:00:00 +0000</pubDate><guid>https://althora.com/blog/fake-ai-tools-malware-small-business/</guid><description>&lt;p&gt;There is a pattern in the 2026 threat data that is worth an office manager&amp;rsquo;s attention, because it targets a habit that spread across small businesses faster than any policy could keep up with: downloading AI tools.&lt;/p&gt;
&lt;p&gt;Kaspersky&amp;rsquo;s SMB threat report for 2026 found &lt;strong&gt;over 33,300 attacks on small and medium businesses in the first four months of the year in which the malware was disguised as a popular AI tool&lt;/strong&gt; — close to five times the figure for the same period in 2025. Researchers counted more than 1,100 distinct malware samples wearing the names of just five well-known AI applications (&lt;a href="https://securelist.com/smb-threat-report-2026/120357/"&gt;Securelist&lt;/a&gt;).&lt;/p&gt;</description></item></channel></rss>