{"id":53,"date":"2026-05-18T12:37:27","date_gmt":"2026-05-18T12:37:27","guid":{"rendered":"https:\/\/convly.ai\/rag-retrieval-augmented-generation-explained\/"},"modified":"2026-06-10T05:06:09","modified_gmt":"2026-06-10T05:06:09","slug":"rag-retrieval-augmented-generation-explained","status":"publish","type":"post","link":"https:\/\/convly.ai\/pt\/rag-retrieval-augmented-generation-explained\/","title":{"rendered":"RAG explicado: como a gera\u00e7\u00e3o aumentada por recupera\u00e7\u00e3o funciona em 2026"},"content":{"rendered":"<p>If you&#8217;ve used an AI tool that answers questions about your company&#8217;s documents, your codebase, or a specific knowledge base, you&#8217;ve used <strong>RAG<\/strong> \u2014 retrieval-augmented generation. It is the single most important architecture pattern in applied AI, and the reason large language models can be useful on information they were never trained on.<\/p>\n<p>This guide explains RAG clearly: what it is, why it exists, how it works step by step, and how to build one. No unnecessary jargon.<\/p>\n<div class=\"convly-tldr\">\n<h3>Principais conclus\u00f5es<\/h3>\n<ul>\n<li><strong>RAG<\/strong> connects a language model to an external knowledge source so it can answer from <em>your<\/em> data.<\/li>\n<li><strong>Why it matters:<\/strong> it fixes the LLM&#8217;s two biggest limits \u2014 outdated knowledge and made-up answers.<\/li>\n<li><strong>How it works:<\/strong> retrieve relevant text, add it to the prompt, then let the model generate an answer grounded in it.<\/li>\n<li><strong>The core tools:<\/strong> embeddings, a vector database, and a retrieval step in front of the model.<\/li>\n<li><strong>RAG vs fine-tuning:<\/strong> RAG adds knowledge; fine-tuning changes behavior. Most projects need RAG first.<\/li>\n<\/ul>\n<\/div>\n<div id=\"ez-toc-container\" class=\"ez-toc-v2_0_84 counter-flat ez-toc-counter ez-toc-container-direction\">\n<label for=\"ez-toc-cssicon-toggle-item-6a38a904dfb0e\" class=\"ez-toc-cssicon-toggle-label\"><span class=\"\"><span class=\"eztoc-hide\" style=\"display:none;\">Alternar<\/span><span class=\"ez-toc-icon-toggle-span\"><svg style=\"fill: #000000;color:#000000\" xmlns=\"http:\/\/www.w3.org\/2000\/svg\" class=\"list-377408\" width=\"20px\" height=\"20px\" viewbox=\"0 0 24 24\" fill=\"none\"><path d=\"M6 6H4v2h2V6zm14 0H8v2h12V6zM4 11h2v2H4v-2zm16 0H8v2h12v-2zM4 16h2v2H4v-2zm16 0H8v2h12v-2z\" fill=\"currentColor\"><\/path><\/svg><svg style=\"fill: #000000;color:#000000\" class=\"arrow-unsorted-368013\" xmlns=\"http:\/\/www.w3.org\/2000\/svg\" width=\"10px\" height=\"10px\" viewbox=\"0 0 24 24\" version=\"1.2\" baseprofile=\"tiny\"><path d=\"M18.2 9.3l-6.2-6.3-6.2 6.3c-.2.2-.3.4-.3.7s.1.5.3.7c.2.2.4.3.7.3h11c.3 0 .5-.1.7-.3.2-.2.3-.5.3-.7s-.1-.5-.3-.7zM5.8 14.7l6.2 6.3 6.2-6.3c.2-.2.3-.5.3-.7s-.1-.5-.3-.7c-.2-.2-.4-.3-.7-.3h-11c-.3 0-.5.1-.7.3-.2.2-.3.5-.3.7s.1.5.3.7z\"\/><\/svg><\/span><\/span><\/label><input type=\"checkbox\"  id=\"ez-toc-cssicon-toggle-item-6a38a904dfb0e\"  aria-label=\"Alternar\" \/><nav><ul class='ez-toc-list ez-toc-list-level-1' ><li class='ez-toc-page-1'><a class=\"ez-toc-link ez-toc-heading-1\" href=\"https:\/\/convly.ai\/pt\/rag-retrieval-augmented-generation-explained\/#The_problem_RAG_solves\" >The problem RAG solves<\/a><\/li><li class='ez-toc-page-1'><a class=\"ez-toc-link ez-toc-heading-2\" href=\"https:\/\/convly.ai\/pt\/rag-retrieval-augmented-generation-explained\/#How_RAG_works_step_by_step\" >How RAG works, step by step<\/a><\/li><li class='ez-toc-page-1'><a class=\"ez-toc-link ez-toc-heading-3\" href=\"https:\/\/convly.ai\/pt\/rag-retrieval-augmented-generation-explained\/#A_simple_analogy\" >A simple analogy<\/a><\/li><li class='ez-toc-page-1'><a class=\"ez-toc-link ez-toc-heading-4\" href=\"https:\/\/convly.ai\/pt\/rag-retrieval-augmented-generation-explained\/#Why_RAG_matters\" >Why RAG matters<\/a><\/li><li class='ez-toc-page-1'><a class=\"ez-toc-link ez-toc-heading-5\" href=\"https:\/\/convly.ai\/pt\/rag-retrieval-augmented-generation-explained\/#What_you_need_to_build_a_RAG_system\" >What you need to build a RAG system<\/a><\/li><li class='ez-toc-page-1'><a class=\"ez-toc-link ez-toc-heading-6\" href=\"https:\/\/convly.ai\/pt\/rag-retrieval-augmented-generation-explained\/#What_makes_RAG_hard_to_do_well\" >What makes RAG hard to do well<\/a><\/li><li class='ez-toc-page-1'><a class=\"ez-toc-link ez-toc-heading-7\" href=\"https:\/\/convly.ai\/pt\/rag-retrieval-augmented-generation-explained\/#When_RAG_is_the_right_tool_and_when_it_isnt\" >When RAG is the right tool (and when it isn&#8217;t)<\/a><\/li><li class='ez-toc-page-1'><a class=\"ez-toc-link ez-toc-heading-8\" href=\"https:\/\/convly.ai\/pt\/rag-retrieval-augmented-generation-explained\/#FAQ\" >Perguntas frequentes<\/a><\/li><li class='ez-toc-page-1'><a class=\"ez-toc-link ez-toc-heading-9\" href=\"https:\/\/convly.ai\/pt\/rag-retrieval-augmented-generation-explained\/#Bottom_line\" >Conclus\u00e3o<\/a><\/li><li class='ez-toc-page-1'><a class=\"ez-toc-link ez-toc-heading-10\" href=\"https:\/\/convly.ai\/pt\/rag-retrieval-augmented-generation-explained\/#Related_articles\" >Artigos relacionados<\/a><\/li><\/ul><\/nav><\/div>\n<h2><span class=\"ez-toc-section\" id=\"The_problem_RAG_solves\"><\/span>The problem RAG solves<span class=\"ez-toc-section-end\"><\/span><\/h2>\n<p>A large language model knows only what it learned during training. That creates two hard limits:<\/p>\n<ol>\n<li><strong>Its knowledge has a cutoff date.<\/strong> It doesn&#8217;t know what happened after training, and it doesn&#8217;t know anything about <em>your<\/em> private documents.<\/li>\n<li><strong>It can hallucinate.<\/strong> Asked something outside its knowledge, an LLM often produces a confident, plausible, wrong answer rather than admitting it doesn&#8217;t know.<\/li>\n<\/ol>\n<p>You could retrain the model on new information, but that&#8217;s slow, expensive, and impractical to do every time a document changes. RAG is the elegant alternative: instead of putting the knowledge <em>inside<\/em> the model, you keep it outside and <strong>hand the model the relevant piece at question time.<\/strong><\/p>\n<h2><span class=\"ez-toc-section\" id=\"How_RAG_works_step_by_step\"><\/span>How RAG works, step by step<span class=\"ez-toc-section-end\"><\/span><\/h2>\n<p>RAG has two phases. The first happens once (or whenever your data changes); the second happens on every question.<\/p>\n<h3>Phase 1: Indexing your knowledge (done ahead of time)<\/h3>\n<ol>\n<li><strong>Collect your documents<\/strong> \u2014 PDFs, web pages, support tickets, code, anything.<\/li>\n<li><strong>Split them into chunks<\/strong> \u2014 break each document into smaller passages, because you want to retrieve precise, relevant snippets, not entire files.<\/li>\n<li><strong>Create embeddings<\/strong> \u2014 pass each chunk through an embedding model, which converts text into a list of numbers (a vector) that captures its meaning. Passages about similar topics end up with similar vectors.<\/li>\n<li><strong>Store them in a vector database<\/strong> \u2014 save every chunk and its vector in a database built for fast similarity search.<\/li>\n<\/ol>\n<h3>Phase 2: Answering a question (done every time)<\/h3>\n<ol>\n<li><strong>Embed the question<\/strong> \u2014 convert the user&#8217;s question into a vector with the same embedding model.<\/li>\n<li><strong>Retrieve<\/strong> \u2014 search the vector database for the chunks whose vectors are most similar to the question&#8217;s vector. These are the passages most likely to contain the answer.<\/li>\n<li><strong>Augment the prompt<\/strong> \u2014 insert those retrieved chunks into the prompt, alongside the question, with an instruction like &#8220;answer using only the context below.&#8221;<\/li>\n<li><strong>Generate<\/strong> \u2014 the LLM writes an answer grounded in the supplied passages, not in its memory.<\/li>\n<\/ol>\n<p>The result: an answer based on <em>your<\/em> current, specific information \u2014 often with citations back to the source chunks.<\/p>\n<h2><span class=\"ez-toc-section\" id=\"A_simple_analogy\"><\/span>A simple analogy<span class=\"ez-toc-section-end\"><\/span><\/h2>\n<p>Think of a plain LLM as a brilliant expert taking a closed-book exam: fluent and knowledgeable, but limited to memory, and prone to bluffing on anything it doesn&#8217;t know.<\/p>\n<p>RAG turns it into an <strong>open-book exam<\/strong>. Before answering each question, the expert is handed the exact pages of the textbook that are relevant. They still need the intelligence to read, synthesize, and explain \u2014 but now the facts come from the book, not from possibly-faulty memory.<\/p>\n<h2><span class=\"ez-toc-section\" id=\"Why_RAG_matters\"><\/span>Why RAG matters<span class=\"ez-toc-section-end\"><\/span><\/h2>\n<p>RAG is the foundation of most useful enterprise AI in 2026:<\/p>\n<ul>\n<li><strong>Grounded answers<\/strong> \u2014 responses are based on real source documents, which sharply reduces hallucination.<\/li>\n<li><strong>Current information<\/strong> \u2014 update the knowledge base and the system instantly &#8220;knows&#8221; the new content; no retraining.<\/li>\n<li><strong>Private data<\/strong> \u2014 it lets a model work with your internal documents without those documents ever being part of model training.<\/li>\n<li><strong>Cita\u00e7\u00f5es<\/strong> \u2014 because you know which chunks were retrieved, you can show users exactly where an answer came from.<\/li>\n<li><strong>Custo<\/strong> \u2014 far cheaper than fine-tuning, and far easier to keep up to date.<\/li>\n<\/ul>\n<p>This is why RAG powers customer-support bots, internal knowledge assistants, documentation search, legal and medical research tools, and &#8220;chat with your codebase&#8221; features.<\/p>\n<h2><span class=\"ez-toc-section\" id=\"What_you_need_to_build_a_RAG_system\"><\/span>What you need to build a RAG system<span class=\"ez-toc-section-end\"><\/span><\/h2>\n<table class=\"convly-vs\">\n<thead>\n<tr>\n<th>Component<\/th>\n<th>Job<\/th>\n<th>Common choices<\/th>\n<\/tr>\n<\/thead>\n<tbody>\n<tr>\n<td>Embedding model<\/td>\n<td>Turn text into meaning-vectors<\/td>\n<td>OpenAI, Cohere, or open-source embedding models<\/td>\n<\/tr>\n<tr>\n<td>Vector database<\/td>\n<td>Store vectors, do fast similarity search<\/td>\n<td>Pinecone, Weaviate, Qdrant, pgvector, Chroma<\/td>\n<\/tr>\n<tr>\n<td>LLM<\/td>\n<td>Generate the final grounded answer<\/td>\n<td>GPT, Claude, Gemini, or an open model<\/td>\n<\/tr>\n<tr>\n<td>Orchestration<\/td>\n<td>Glue the steps together<\/td>\n<td>LangChain, LlamaIndex, or custom code<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<p>A basic RAG prototype can be built in an afternoon. A <em>good<\/em> production RAG system is harder \u2014 the quality lives in the details below.<\/p>\n<h2><span class=\"ez-toc-section\" id=\"What_makes_RAG_hard_to_do_well\"><\/span>What makes RAG hard to do well<span class=\"ez-toc-section-end\"><\/span><\/h2>\n<p>A naive RAG system works in a demo and disappoints in production. The difficult parts:<\/p>\n<ul>\n<li><strong>Chunking strategy<\/strong> \u2014 chunks too large bury the answer in noise; too small and they lose context. Getting this right matters more than people expect.<\/li>\n<li><strong>Retrieval quality<\/strong> \u2014 if the retrieval step fetches the wrong passages, the LLM cannot save you. &#8220;Garbage in, garbage out&#8221; is the central RAG failure mode.<\/li>\n<li><strong>Hybrid search<\/strong> \u2014 pure vector similarity misses exact keywords, names, and codes; the best systems combine vector search with traditional keyword search.<\/li>\n<li><strong>Reranking<\/strong> \u2014 a second model that re-scores retrieved chunks for relevance noticeably improves answer quality.<\/li>\n<li><strong>Evaluation<\/strong> \u2014 you need a way to measure whether retrieval and answers are actually good, not just &#8220;looks fine.&#8221;<\/li>\n<\/ul>\n<p>The phrase to remember: in RAG, <strong>retrieval quality is the ceiling on answer quality.<\/strong><\/p>\n<p><!--ai-enriched--><\/p>\n<h2><span class=\"ez-toc-section\" id=\"When_RAG_is_the_right_tool_and_when_it_isnt\"><\/span>When RAG is the right tool (and when it isn&#8217;t)<span class=\"ez-toc-section-end\"><\/span><\/h2>\n<p>RAG is not the answer to every problem. Modern frontier models now accept enormous contexts \u2014 Gemini, Llama 4, and Grok advertise windows measured in the millions of tokens \u2014 which means you can sometimes just paste the whole document into the prompt and skip retrieval entirely. The smart question in 2026 is not &#8220;RAG or no RAG?&#8221; but &#8220;where does the knowledge live, and how much of it is relevant to any one question?&#8221;<\/p>\n<p>Reach for RAG when the knowledge is large, changing, and only partly relevant to each query: a 10,000-page documentation set, a support knowledge base, a legal or compliance archive, an internal wiki. Retrieval pulls the handful of passages that actually matter, which keeps answers grounded, cuts token cost dramatically, and lets you attach citations so a reader can verify the source. It also updates instantly \u2014 add a document to the index and the system knows it on the next question, with no retraining.<\/p>\n<p>Lean on a long-context model instead when the relevant material is small enough to fit comfortably and the task needs the <strong>whole<\/strong> document at once \u2014 summarizing a single contract, reasoning across an entire codebase, or answering questions that depend on connections scattered throughout one file. Stuffing everything into the prompt also has a quiet failure mode: models reliably lose accuracy when the key fact sits buried in the middle of a very long context rather than near the start or end. Retrieval sidesteps that by surfacing the right passage directly.<\/p>\n<p>A simple rule of thumb:<\/p>\n<ul>\n<li><strong>Big corpus, narrow questions<\/strong> (each answer needs a small slice) \u2014 use RAG.<\/li>\n<li><strong>One document, holistic question<\/strong> (the answer needs all of it) \u2014 use long context.<\/li>\n<li><strong>Source attribution is mandatory<\/strong> (legal, finance, healthcare, anything audited) \u2014 use RAG; citations are the point.<\/li>\n<li><strong>Knowledge changes daily<\/strong> \u2014 use RAG, so updates are a data operation, not a model operation.<\/li>\n<\/ul>\n<p>The highest-ROI deployments cluster where these conditions overlap: customer support assistants grounded in current product docs, internal knowledge search, compliance and policy lookup, and financial or technical research. In practice the strongest systems blend both approaches \u2014 RAG to find the right material, a capable long-context model to reason over it.<\/p>\n<h2><span class=\"ez-toc-section\" id=\"FAQ\"><\/span>Perguntas frequentes<span class=\"ez-toc-section-end\"><\/span><\/h2>\n<h3>What is RAG in simple terms?<\/h3>\n<p>RAG (retrieval-augmented generation) is a technique that lets an AI model answer questions using external information instead of only its training data. It retrieves relevant passages from a knowledge source and gives them to the model, so the answer is grounded in real, specific documents.<\/p>\n<h3>Why is RAG better than just asking the LLM directly?<\/h3>\n<p>A plain LLM only knows its training data, which is fixed and has a cutoff date \u2014 and it can confidently make things up. RAG supplies current, specific, private information at question time, so answers are accurate, up to date, and traceable to a source.<\/p>\n<h3>What is the difference between RAG and fine-tuning?<\/h3>\n<p>RAG <em>adds knowledge<\/em> by retrieving documents at question time; fine-tuning <em>changes behavior<\/em> by further training the model on examples. RAG is the right tool when the model needs facts it doesn&#8217;t have; fine-tuning is right for teaching a style, format, or task. They can be combined.<\/p>\n<h3>Preciso de um banco de dados vetorial para RAG?<\/h3>\n<p>For anything beyond a tiny prototype, yes. A vector database stores the meaning-vectors of your text chunks and performs fast similarity search to find relevant passages. Options range from managed services to libraries and the pgvector extension for PostgreSQL.<\/p>\n<h3>Does RAG eliminate hallucinations?<\/h3>\n<p>It greatly reduces them, but doesn&#8217;t eliminate them. If retrieval fetches the right passages and the prompt instructs the model to answer only from them, hallucination drops sharply. But poor retrieval, or a model ignoring the context, can still produce errors \u2014 which is why retrieval quality and evaluation matter.<\/p>\n<h3>Is RAG still necessary now that models have million-token context windows?<\/h3>\n<p>Often, yes. Huge context windows let you skip retrieval for a single document, but they do not solve large, fast-changing knowledge bases. Feeding millions of tokens on every query is slow and expensive, and accuracy drops when the key fact is buried deep in the context. RAG retrieves only the relevant passages, so it stays cheaper, faster, and more current \u2014 and it gives you citations. The two are complementary, not rivals.<\/p>\n<h3>How do I get a RAG system to cite its sources?<\/h3>\n<p>Citations are one of RAG&#8217;s biggest practical advantages, but you have to build them in. Store metadata \u2014 document title, URL, page or section \u2014 alongside each chunk when you index. At answer time, pass the retrieved chunks to the model with their identifiers and instruct it to reference which source each claim came from. Your application then links those identifiers back to the original documents, so a reader can verify every statement.<\/p>\n<h3>How do I keep a RAG system&#8217;s knowledge up to date?<\/h3>\n<p>This is RAG&#8217;s structural strength over fine-tuning: updating knowledge is a data task, not a training task. When a source document changes, you re-chunk and re-embed just that document and refresh its entries in the vector database; the rest of the index is untouched. Most production systems run this on a schedule or trigger it whenever a source file is added or edited, so the assistant reflects the latest information within minutes rather than waiting for a model retrain.<\/p>\n<h2><span class=\"ez-toc-section\" id=\"Bottom_line\"><\/span>Conclus\u00e3o<span class=\"ez-toc-section-end\"><\/span><\/h2>\n<p>RAG is the bridge between a general-purpose language model and <em>your<\/em> specific, current, private knowledge. It works by retrieving the relevant text and handing it to the model at question time \u2014 turning a closed-book exam into an open-book one.<\/p>\n<p>It is the default architecture for almost every serious enterprise AI application in 2026, and the first thing to reach for when you need an AI that answers from your own data. A basic version is quick to build; a great one depends on getting chunking, retrieval, and evaluation right. If you&#8217;re choosing between RAG and fine-tuning, start with RAG \u2014 our <a href=\"\/pt\/fine-tuning-vs-rag\/\">fine-tuning vs RAG guide<\/a> explains exactly when you need each.<\/p>\n<p><!--related-block--><\/p>\n<div class=\"convly-related\">\n<h2><span class=\"ez-toc-section\" id=\"Related_articles\"><\/span>Artigos relacionados<span class=\"ez-toc-section-end\"><\/span><\/h2>\n<ul>\n<li><a href=\"https:\/\/convly.ai\/pt\/claude-5-new-ai-models-june-2026\/\">Existe um Claude 5? Claude Fable 5 e todos os principais modelos de IA de junho de 2026<\/a><\/li>\n<li><a href=\"https:\/\/convly.ai\/pt\/llm-hallucinations-complete-guide\/\">Alucina\u00e7\u00f5es de LLMs em 2026: Por que ocorrem e como evit\u00e1-las<\/a><\/li>\n<li><a href=\"https:\/\/convly.ai\/pt\/prompt-engineering-techniques\/\">Engenharia de prompts em 2026: 12 t\u00e9cnicas que realmente funcionam<\/a><\/li>\n<li><a href=\"https:\/\/convly.ai\/pt\/what-is-ollama-complete-guide-2026\/\">O que \u00e9 o Ollama? Guia completo para executar LLMs localmente em 2026<\/a><\/li>\n<\/ul>\n<\/div>","protected":false},"excerpt":{"rendered":"<p>RAG is the technique behind almost every AI system that answers questions from your own documents. This guide explains how retrieval-augmented generation works \u2014 clearly, and without the jargon.<\/p>","protected":false},"author":0,"featured_media":54,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"site-sidebar-layout":"default","site-content-layout":"","ast-site-content-layout":"default","site-content-style":"default","site-sidebar-style":"default","ast-global-header-display":"","ast-banner-title-visibility":"","ast-main-header-display":"","ast-hfb-above-header-display":"","ast-hfb-below-header-display":"","ast-hfb-mobile-header-display":"","site-post-title":"","ast-breadcrumbs-content":"","ast-featured-img":"","footer-sml-layout":"","ast-disable-related-posts":"","theme-transparent-header-meta":"","adv-header-id-meta":"","stick-header-meta":"","header-above-stick-meta":"","header-main-stick-meta":"","header-below-stick-meta":"","astra-migrate-meta-layouts":"default","ast-page-background-enabled":"default","ast-page-background-meta":{"desktop":{"background-color":"var(--ast-global-color-5)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"tablet":{"background-color":"","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"mobile":{"background-color":"","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""}},"ast-content-background-meta":{"desktop":{"background-color":"var(--ast-global-color-4)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"tablet":{"background-color":"var(--ast-global-color-4)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"mobile":{"background-color":"var(--ast-global-color-4)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""}},"footnotes":""},"categories":[3],"tags":[431,442,429,440,441],"class_list":["post-53","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-llms","tag-ai-architecture","tag-llm","tag-rag","tag-retrieval-augmented-generation","tag-vector-database"],"_links":{"self":[{"href":"https:\/\/convly.ai\/pt\/wp-json\/wp\/v2\/posts\/53","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/convly.ai\/pt\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/convly.ai\/pt\/wp-json\/wp\/v2\/types\/post"}],"replies":[{"embeddable":true,"href":"https:\/\/convly.ai\/pt\/wp-json\/wp\/v2\/comments?post=53"}],"version-history":[{"count":3,"href":"https:\/\/convly.ai\/pt\/wp-json\/wp\/v2\/posts\/53\/revisions"}],"predecessor-version":[{"id":1045,"href":"https:\/\/convly.ai\/pt\/wp-json\/wp\/v2\/posts\/53\/revisions\/1045"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/convly.ai\/pt\/wp-json\/wp\/v2\/media\/54"}],"wp:attachment":[{"href":"https:\/\/convly.ai\/pt\/wp-json\/wp\/v2\/media?parent=53"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/convly.ai\/pt\/wp-json\/wp\/v2\/categories?post=53"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/convly.ai\/pt\/wp-json\/wp\/v2\/tags?post=53"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}