{"id":14645,"date":"2026-05-04T11:09:12","date_gmt":"2026-05-04T10:09:12","guid":{"rendered":"https:\/\/www.thinkupc.com\/?p=14645"},"modified":"2026-05-07T09:38:33","modified_gmt":"2026-05-07T08:38:33","slug":"ai-is-a-7-story-building","status":"publish","type":"post","link":"https:\/\/www.thinkupc.com\/en\/news\/ai-is-a-7-story-building\/","title":{"rendered":"AI Is a 7-Story Building"},"content":{"rendered":"\t\t<div data-elementor-type=\"wp-post\" data-elementor-id=\"14645\" class=\"elementor elementor-14645 elementor-14604\" data-elementor-post-type=\"post\">\n\t\t\t\t<div class=\"elementor-element elementor-element-6772d2b e-flex e-con-boxed e-con e-parent\" data-id=\"6772d2b\" data-element_type=\"container\" data-e-type=\"container\">\n\t\t\t\t\t<div class=\"e-con-inner\">\n\t\t\t\t<div class=\"elementor-element elementor-element-3b78d3d elementor-widget elementor-widget-text-editor\" data-id=\"3b78d3d\" data-element_type=\"widget\" data-e-type=\"widget\" data-widget_type=\"text-editor.default\">\n\t\t\t\t\t\t\t\t\t<p>Back in 2017, NVIDIA&#8217;s CEO Jensen Huang stated in an interview: &#8220;<a href=\"https:\/\/www.technologyreview.com\/2017\/05\/12\/151722\/nvidia-ceo-software-is-eating-the-world-but-ai-is-going-to-eat-software\/\">AI is going to eat software<\/a>&#8221; He did so drawing inspiration from \u2014or paying homage to\u2014 the title of a famous article by Marc Andreessen, published in the Wall Street Journal six years earlier, in which he explained <a href=\"https:\/\/a16z.com\/why-software-is-eating-the-world\/\">why software was eating the world<\/a>.<\/p>\n<p>These articles are part of the recent history of AI, and I often use them to explain how we got to where we are today. But that&#8217;s not what I&#8217;ll be discussing in today&#8217;s article. What I want to review are some<strong> important challenges that all companies and organizations<\/strong> face in light of the accelerated revolution being driven by generative, reasoning, and agentic AI; and I&#8217;ll do so by drawing inspiration from the title of a <a href=\"https:\/\/blogs.nvidia.com\/blog\/ai-5-layer-cake\/\">recent article published by Jensen Huang himself<\/a>. Since he&#8217;s also fond of homages, I&#8217;m sure he won&#8217;t mind.<\/p>\n<h2>AI Is a 7-Story Building<\/h2>\n<p>If we look at the components that make up the traditional structure of our organizations&#8217; information systems, we essentially have\u2026<\/p>\n<ul>\n<li><strong>\u2026applications<\/strong>, oriented toward different user profiles and working with\u2026<\/li>\n<li><strong>\u2026data<\/strong>, internal or external to our organization, running on\u2026<\/li>\n<li><strong>\u2026infrastructure<\/strong>, for storage, networking, or computing, making use of\u2026<\/li>\n<li><strong>\u2026chips<\/strong>, designed to execute mathematical operations consuming\u2026<\/li>\n<li><strong>\u2026energy<\/strong>, as efficiently as possible.<\/li>\n<\/ul>\n<p>We could imagine these 5 levels as 5 stories of a building, where each story needs to rest robustly on the ones below in order to deliver real value to the user on the 5th floor.<\/p>\n<p>Well, since AI is changing everything, I believe it&#8217;s also substantially transforming this building. In fact, to take full advantage of agentic AI, I think we need to construct \u2014at least mentally\u2014 a new building next to it, with 7 stories. Let&#8217;s draw it.<\/p>\n<p>And now we can ask ourselves:<strong> is it really a new building?<\/strong><\/p>\n<p>I think it is, because the legacy building, where traditional deterministic applications live, isn&#8217;t going away; we&#8217;ll keep it and we&#8217;ll have to maintain it for many years to come.<\/p>\n<p>The new building, on the other hand, where agentic AI will live, has quite different foundations. Let&#8217;s briefly review its characteristics, story by story.<\/p>\n<h3>1. Energy<\/h3>\n<p>It&#8217;s often said that AI is, ultimately, electricity transformed into computation. It sounds obvious, and one might think traditional applications are exactly the same. But there&#8217;s such a huge difference in scale that this layer must be understood as the foundation of something new.<\/p>\n<p>A few months ago, OpenAI ended 2025 operating with an <a href=\"https:\/\/fanaticosdelhardware.com\/openai-apunta-a-30gw-de-computacion-en-2030-un-reto-que-pone-al-limite-a-la-industria-de-semiconductores\/\">energy capacity of 2 GW<\/a> to run ChatGPT. That&#8217;s the equivalent of 2 nuclear power plants, or the needs of a city like Barcelona. According to the company itself, <a href=\"https:\/\/www.tomshardware.com\/tech-industry\/artificial-intelligence\/openais-colossal-ai-data-center-targets-would-consume-as-much-electricity-as-entire-nation-of-india-250gw-target-would-require-30-million-gpus-annually-to-ensure-continuous-operation-emit-twice-as-much-carbon-dioxide-as-exxonmobil\">by 2033 they will need 250 GW<\/a>, the equivalent of 250 nuclear reactors. That figure means that a single company will require more than half of the installed capacity of the 437 reactors currently operating across the entire planet.<\/p>\n<p>This shift in scale, with clear <strong>implications for sustainability and availability<\/strong>, will lead many organizations to require new talent capable of analyzing and optimizing the energy footprint of their AI solutions. For other companies, the strategic decision will be clear: delegate the cost of energy to the provider via price per token and focus on the upper floors. But delegating doesn&#8217;t mean ignoring; we&#8217;ll need to know what kind of energy is powering our agents, because more and more clients, regulators, and investors will be asking. In the legacy building, this topic was barely relevant. In the new building, it&#8217;s a top priority.<\/p>\n<h3>2. Chips<\/h3>\n<p>If energy is the fuel, chips are the engine. And in recent years, they&#8217;ve stopped being just another component to become the real bottleneck of the AI economy. A single H100 GPU costs around 30,000 dollars, and the training centers for <a href=\"https:\/\/www.frontiermodelforum.org\/about-us\/\">frontier models<\/a> use tens or hundreds of thousands of them. Demand far exceeds supply, and access often depends more on your relationship with the provider than on your available budget. In the legacy building, servers were a commodity you bought from a catalog. In the new building, chips are a scarce and strategic asset.<\/p>\n<p>The category, moreover, has diversified: general-purpose GPUs (the ubiquitous NVIDIAs), proprietary accelerators from the hyperscalers (Google&#8217;s TPUs, AWS&#8217;s Trainium, Microsoft&#8217;s Maia), inference-specialized chips with extremely low latency (Groq, Cerebras), and embedded NPUs to run small models on phones or laptops. All of this depending on a very short supply chain: TSMC and Samsung manufacturing the chips, and ASML producing the EUV lithography machines. Pure geopolitics.<\/p>\n<p>The good news is that <a href=\"https:\/\/ecosistemastartup.com\/google-tpu-8a-gen-dos-chips-para-la-era-agentica\/\">a healthy separation is starting to emerge<\/a> between training silicon \u2014extremely expensive, with few players\u2014 and inference silicon, where the ecosystem is more open and competitive, and which is what will enable real-time agents at reasonable cost. And here a question arises that just a few years ago wasn&#8217;t on the agenda of any board of directors: <strong>technological sovereignty<\/strong>. Organizations \u2014and countries\u2014 that want to retain real decision-making capacity in the new AI building will need to bring in new talent that is currently scarce: architecture engineers and specialists capable of squeezing every watt and every cycle out of the available hardware. In the legacy building, chips were a procurement decision. In the new building, they are a decision about the future.<\/p>\n<h3>3. Infrastructure<\/h3>\n<p>Story 3 is the factory that converts chips and energy into consumable capacity. It allows, for example, a ten-person startup to access the same compute power as a large bank, simply on a pay-as-you-go basis.<\/p>\n<p>Here we find the classic hyperscalers, with services like AWS Bedrock, Azure AI Foundry, or Google Vertex AI, and the AI-specialized neoclouds. On top of these providers, we build training clusters, inference platforms, AI gateways that centralize routing and policies, and the entire MLOps\/LLMOps discipline: CI\/CD, observability, prompt versioning, and continuous evaluation.<\/p>\n<p>The critical decisions on this story are mostly about <strong>governance<\/strong>:<\/p>\n<ul>\n<li>Which workloads can go to public APIs and which require private, on-premises deployment?<\/li>\n<li>In which regions are we willing to operate, especially when dealing with sensitive or regulated data?<\/li>\n<li>What model of data sovereignty do we guarantee to our clients?<\/li>\n<li>Which platforms do we standardize internally to prevent the proliferation of <em>shadow AI<\/em>? I&#8217;m referring to that very common phenomenon in which each team ends up assembling its own toolkit outside the organization&#8217;s umbrella, with all the security, cost, and compliance risks that entails.<\/li>\n<\/ul>\n<p>New talent will be needed here too. In the legacy building, platform teams are very skilled in networking, virtualization, and containers. In the new building, what&#8217;s needed are profiles who master GPU orchestration, distributed inference architectures, and agent observability, capable of evaluating performance from every angle: cost, latency, quality, and alignment.<\/p>\n<h3>4. Data<\/h3>\n<p>We arrive at the story where real differentiation begins. General-purpose models are, to a large extent, a commodity accessible via API. But each organization&#8217;s own knowledge \u2014its contracts, its customer relationships, its technical documentation, its processes\u2014 is not, and never will be.<\/p>\n<p>In the legacy building, we already had data, well structured in relational databases, data warehouses, and data lakes. All of this is still valid. But the new building also requires being able to work with <strong>unstructured data at scale<\/strong> (PDFs, tickets, transcripts, images, audio\u2026) and turning it into something that a model \u2014on the floor above\u2014 can understand and use in a reliable, traceable, and governed way. New elements emerge, such as <a href=\"https:\/\/x.com\/karpathy\/status\/2039805659525644595\">AI-specific ingestion and preparation<\/a>, vector databases that support semantic search and RAG, and knowledge graphs. Agents will consume this data in various ways: via RAG, via <em>tool use<\/em> on structured data, with light fine-tuning when the domain calls for it, or via the agent&#8217;s own memory, capitalizing on past conversations and decisions to improve future ones.<\/p>\n<p>But the most relevant aspect of this story is, once again, governance: the catalog, the lineage from the source to the prompt, the classification (public, internal, confidential, personal), GDPR compliance, anonymization, and so on. An agent is only as good and as safe as the data it accesses. The key idea on this story, and probably of the entire building, is this:<strong> lasting competitive advantage in AI almost never lies in the model; it lies in your own quality data, well governed and well connected<\/strong>.<\/p>\n<p>In the legacy building, data is an asset that gets queried. In the new building, it is an asset that reasons, alongside AI agents and ourselves.<\/p>\n<h3>5. Models<\/h3>\n<p>If data is the raw material, models are the brain of the new building. And I think it&#8217;s important to enter this story with a key first idea: it no longer makes sense to talk about &#8220;choosing the best LLM or the best provider&#8221;. What you have to choose is a portfolio of models depending on the task, cost, latency, privacy, and reasoning capability. We have:<\/p>\n<ul>\n<li><strong>Frontier models<\/strong> (such as Gemini, GPT, or Claude), high-performance, accessible via API, and typically multimodal, combining text, image, audio, and video.<\/li>\n<li><strong>Extended-reasoning models<\/strong>, optimized for planning or coding, suitable for agents that chain multiple steps.<\/li>\n<li><strong>Open-weights models<\/strong> (such as Mistral, DeepSeek, Qwen, or Gemma) that allow private deployment, fine-tuning, and sovereignty.<\/li>\n<li><strong>Small models<\/strong> (SLMs), fast and cheap, ideal for specific tasks like classification, extraction, or routing within an agent.<\/li>\n<li>A good number of <strong>specialized models<\/strong>: embeddings, OCR, speech-to-text, image or code generation.<\/li>\n<li>And the <strong>classic ML models<\/strong> \u2014forecasting, anomaly detection, scoring\u2014 which haven&#8217;t disappeared and live alongside the generative ones inside the same agents.<\/li>\n<\/ul>\n<p>Managing this story well will require choosing the right model for each task \u2014not all of them need a frontier LLM\u2014, doing multi-model routing for cost and capability, and building your own evaluations based on your actual use cases. Public benchmarks are indicative; the only ones that really matter are those that measure whether the model does the job I need it to do at my company.<\/p>\n<p>But the most important architectural decision is a different one: <strong>maintaining independence from any specific model<\/strong>. Models are updated frequently, and what is state-of-the-art today will be surpassed tomorrow. A good abstraction layer must be created that makes it possible to switch providers, mix open and closed models, and take advantage of each new generation without having to rebuild the applications. It is, arguably, one of the best architectural investments that can be made today.<\/p>\n<h3>6. Applications<\/h3>\n<p>This is the story where agentic AI becomes tangible. It is where a model that generates text becomes an agent that observes, plans, acts, learns, and is accountable. And it must be clear that to design an agentic system, most of the work \u2014and most of the value\u2014 lies in the components surrounding the model, not in the model itself.<\/p>\n<p>In short, a modern agentic system has eight pieces:<\/p>\n<ul>\n<li>A <strong>persona <\/strong>that defines the agent&#8217;s identity: its role, its tone, and the limits of what it can and cannot do. It is, in a way, the agent&#8217;s ID card and the first design decision.<\/li>\n<li>An <strong>orchestrator <\/strong>that decides at each step whether to call the model, use a tool, ask the user for confirmation, or consider the task complete (LangGraph, Semantic Kernel, CrewAI, AutoGen\u2026).<\/li>\n<li>A <strong>multi-layered memory<\/strong>: short-term (the ongoing conversation), long-term (persistent preferences and facts), episodic (reusable past executions), and semantic (stable domain knowledge).<\/li>\n<li>A <strong>set of tools<\/strong> that act as the agent&#8217;s &#8220;hands&#8221; \u2014APIs, code execution, web browsing\u2014 today often connected via the MCP standard.<\/li>\n<li>A<strong> reasoning loop<\/strong> to break objectives down into sub-tasks and re-plan when something goes wrong. And, when needed, multi-agent patterns with roles (a researcher, a writer, a reviewer, and a coordinator).<\/li>\n<li>An <strong>interface <\/strong>that involves humans when needed (<em>human-in-the-loop<\/em>).<\/li>\n<li>An <strong>observability <\/strong>and evaluation layer with full traces of every execution (prompts, tool calls, tokens, cost, latency) to ensure quality.<\/li>\n<li>A <strong>security, compliance, and control layer<\/strong> where the agent is assigned permissions and limits (in budget and time), and where audit systems and input\/output guardrails are implemented.<\/li>\n<\/ul>\n<p>In the legacy building, applications are deterministic: the same input always produces the same output, and quality is validated with unit tests. In the new building, applications are probabilistic: the same input can produce different responses, and quality is measured statistically through continuous evaluation. This shift has profound implications for how applications are built, tested, deployed, and maintained.<\/p>\n<p>In the legacy building, development teams master classic frameworks, microservices architectures, and DevOps practices. The new building also requires profiles with expertise in agent design, orchestration, context engineering, MCP integration, agent observability, and security specific to probabilistic systems.<br \/>In the legacy building, applications execute. In the new building, applications decide.<\/p>\n<h3>7. Use Cases<\/h3>\n<p>We&#8217;ve reached the top story, which is probably the only one that really matters to the business. All the floors below only make sense if they translate into concrete results: productivity, revenue, customer satisfaction, or new products. And it&#8217;s also the most human story of the building: this is where people, processes, culture, and change management coexist.<\/p>\n<p>We will have two major types of use cases. <strong>Passive AI<\/strong> \u2014chatbots that respond or assist\u2014 is enormously useful and often the first thing an organization adopts. <strong>Active AI<\/strong> \u2014agents that plan and execute tasks end-to-end\u2014 shifts our human role toward that of supervisors who review and approve. Active AI requires more process design, permissions, and auditing, but it&#8217;s where the deepest transformations and the greatest value will be unlocked.<\/p>\n<p>This new way of interacting with technology demands a <a href=\"https:\/\/www.thinkupc.com\/en\/news\/designing-in-the-age-of-ai-the-new-role-of-user-experience\/\">new way of designing the user experience (UX)<\/a>. <strong>AI-Native applications<\/strong> are not traditional applications with a chatbot stuck in the corner; they are something different, with their own design principles: <strong>intent over navigation<\/strong> through menus (the user states what they want to achieve, rather than looking for the button that does it); <strong>human control<\/strong>, with the ability to stop, correct, or approve step by step depending on the risk; <strong>transparency <\/strong>about what the agent will do and why;<strong> in-context feedback<\/strong> that allows correction without breaking the flow; real multimodality (text, voice, image, documents); <strong>generative interfaces<\/strong> built dynamically depending on the task; and <strong>reversibility <\/strong>of actions whenever possible.<\/p>\n<p>But there is one differentiating factor on this story that I find even more important than all of the above: users themselves identify, build, and share their own use cases. <strong>users themselves identify, build, and share their own use cases.<\/strong><\/p>\n<p>The era in which an IT department designed all use cases is over. With the right tools, anyone can identify a repetitive task of theirs, build a mini-agent to automate it, and share it with others. This shift is as significant as the one brought about by the spreadsheet in the eighties: it moves the ability to automate work toward the people who know that work best. For this to succeed, the organization must put three conditions on the table:<\/p>\n<ol>\n<li>A<strong> secure corporate platform<\/strong> to build use cases on, where appropriate features, identity, and permissions are applied.<\/li>\n<li>A <strong>marketplace or internal portal<\/strong> of assistants and agents, where useful use cases are discovered, shared, and reused. The organization that best deploys and cultivates this environment gains a huge advantage.<\/li>\n<li>A<strong> lightweight governance<\/strong> that distinguishes what is sensitive (clients, money, personal data) and requires formal approval, from the rest, where it&#8217;s better to leave room for exploration. All of this, accompanied by training in agentic thinking and the sharing of best practices, because AI adoption is, above all else, a phenomenon that has to settle into the corporate culture.<\/li>\n<\/ol>\n<p>New talent will also be needed here, but of a different kind from the lower stories. Not so much new technical profiles, but business people capable of thinking agentically: breaking down processes, defining clear objectives, designing flows with human supervision, measuring results, and iterating. AI product managers, AI-Native UX designers, and new figures such as AI Champions will also be required to drive the rest of the organization forward.<\/p>\n<p>In the legacy building, use cases were specified by the business and implemented by IT. In the new building, use cases are discovered, prototyped, and shared by the very people who do the work.<\/p>\n<h2>So how do we use this building?<\/h2>\n<p>The new 7-story building is not just a descriptive metaphor. The aim of this article is to provide a tool for reflection and decision-making that allows any executive committee to answer four questions, in order:<\/p>\n<ol>\n<li><strong>On which floor are we really competing?<\/strong> For the vast majority of companies, stories 1, 2, and 5 are not the playing field; they are a cost delegated to providers. Infrastructure (story 3) isn&#8217;t the competitive arena for many either, but in some cases, when sovereignty or confidentiality demand it, it will need to be managed on-premises.<\/li>\n<li><strong>Where do we have a differentiating advantage?<\/strong> In our own data (story 4), in the use cases specific to our business (story 7), and in the skill with which we connect both through agents (story 6). This will be the &#8220;secret sauce&#8221; of our business and is where we should focus our efforts.<\/li>\n<li><strong>What are we delegating, and to whom?<\/strong> Energy, to the system. Infrastructure and models, to leading providers (with a multi-model strategy). But never the data or the use cases: those stay at home and aren&#8217;t delegated. We&#8217;ll only need internal and external talent, with trusted experts to support us.<\/li>\n<li><strong>Are we managing the risks well?<\/strong> As we&#8217;ve seen, the new building demands the new talent we just discussed and also brings new risks to manage, from the energy sustainability of story 1 to security or ethical issues in the use cases on story 7. Once again, we&#8217;ll need trusted experts at our side to address them.<\/li>\n<\/ol>\n<p>To sum up, the idea I&#8217;d like to leave you with is this:<\/p>\n<div style=\"margin: 0px; background-color: #f5f5f5; border-radius: 0px; position: relative; padding: 1.25rem 2.5rem 1.25rem 1.5rem;\">\n<p>AI is a 7-story building; the lower floors are bought, the middle ones are designed following the recommendations of experts, and the upper ones are built tailor-made with the people of the organization. The value lies on the upper floors, but without solid foundations they won&#8217;t stand.<\/p>\n<p><strong>And, above all, no building stands on its own. It only stands by bringing together people who know how to build it with people who want to live in it. <\/strong><\/p>\n<\/div>\n\t\t\t\t\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t","protected":false},"excerpt":{"rendered":"<p>Javier Otero reflects on the challenges of agentic AI and the structure companies need to lead the digital revolution.<\/p>\n","protected":false},"author":30,"featured_media":14891,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"_acf_changed":false,"footnotes":""},"categories":[278,256],"tags":[209],"class_list":["post-14645","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-digital-strategy-en","category-news","tag-digital-strategy-en"],"acf":[],"yoast_head":"<!-- This site is optimized with the Yoast SEO plugin v27.5 - https:\/\/yoast.com\/product\/yoast-seo-wordpress\/ -->\n<title>AI Is a 7-Story Building - ThinkUPC<\/title>\n<meta name=\"description\" content=\"Javier Otero reflects on the challenges of agentic AI and the structure companies need to lead the digital revolution.\" \/>\n<meta name=\"robots\" content=\"index, follow, max-snippet:-1, max-image-preview:large, max-video-preview:-1\" \/>\n<link rel=\"canonical\" href=\"https:\/\/www.thinkupc.com\/en\/news\/ai-is-a-7-story-building\/\" \/>\n<meta property=\"og:locale\" content=\"en_US\" \/>\n<meta property=\"og:type\" content=\"article\" \/>\n<meta property=\"og:title\" content=\"AI Is a 7-Story Building - ThinkUPC\" \/>\n<meta property=\"og:description\" content=\"Javier Otero reflects on the challenges of agentic AI and the structure companies need to lead the digital revolution.\" \/>\n<meta property=\"og:url\" content=\"https:\/\/www.thinkupc.com\/en\/news\/ai-is-a-7-story-building\/\" \/>\n<meta property=\"og:site_name\" content=\"ThinkUPC\" \/>\n<meta property=\"article:published_time\" content=\"2026-05-04T10:09:12+00:00\" \/>\n<meta property=\"article:modified_time\" content=\"2026-05-07T08:38:33+00:00\" \/>\n<meta property=\"og:image\" content=\"https:\/\/www.thinkupc.com\/wp-content\/uploads\/2026\/05\/edificiIAI-retall1-noticia.jpg\" \/>\n\t<meta property=\"og:image:width\" content=\"1400\" \/>\n\t<meta property=\"og:image:height\" content=\"683\" \/>\n\t<meta property=\"og:image:type\" content=\"image\/jpeg\" \/>\n<meta name=\"author\" content=\"Javier Otero\" \/>\n<meta name=\"twitter:card\" content=\"summary_large_image\" \/>\n<meta name=\"twitter:label1\" content=\"Written by\" \/>\n\t<meta name=\"twitter:data1\" content=\"Javier Otero\" \/>\n\t<meta name=\"twitter:label2\" content=\"Est. reading time\" \/>\n\t<meta name=\"twitter:data2\" content=\"14 minutes\" \/>\n<script type=\"application\/ld+json\" class=\"yoast-schema-graph\">{\"@context\":\"https:\\\/\\\/schema.org\",\"@graph\":[{\"@type\":\"Article\",\"@id\":\"https:\\\/\\\/www.thinkupc.com\\\/en\\\/news\\\/ai-is-a-7-story-building\\\/#article\",\"isPartOf\":{\"@id\":\"https:\\\/\\\/www.thinkupc.com\\\/en\\\/news\\\/ai-is-a-7-story-building\\\/\"},\"author\":{\"name\":\"Javier Otero\",\"@id\":\"https:\\\/\\\/www.thinkupc.com\\\/en\\\/#\\\/schema\\\/person\\\/e9c4eb2ebe12f20d120c4e4513e613c7\"},\"headline\":\"AI Is a 7-Story Building\",\"datePublished\":\"2026-05-04T10:09:12+00:00\",\"dateModified\":\"2026-05-07T08:38:33+00:00\",\"mainEntityOfPage\":{\"@id\":\"https:\\\/\\\/www.thinkupc.com\\\/en\\\/news\\\/ai-is-a-7-story-building\\\/\"},\"wordCount\":3028,\"publisher\":{\"@id\":\"https:\\\/\\\/www.thinkupc.com\\\/en\\\/#organization\"},\"image\":{\"@id\":\"https:\\\/\\\/www.thinkupc.com\\\/en\\\/news\\\/ai-is-a-7-story-building\\\/#primaryimage\"},\"thumbnailUrl\":\"https:\\\/\\\/www.thinkupc.com\\\/wp-content\\\/uploads\\\/2026\\\/05\\\/edificiIAI-retall1-noticia.jpg\",\"keywords\":[\"Digital Strategy\"],\"articleSection\":[\"Digital Strategy\",\"News\"],\"inLanguage\":\"en-US\"},{\"@type\":\"WebPage\",\"@id\":\"https:\\\/\\\/www.thinkupc.com\\\/en\\\/news\\\/ai-is-a-7-story-building\\\/\",\"url\":\"https:\\\/\\\/www.thinkupc.com\\\/en\\\/news\\\/ai-is-a-7-story-building\\\/\",\"name\":\"AI Is a 7-Story Building - ThinkUPC\",\"isPartOf\":{\"@id\":\"https:\\\/\\\/www.thinkupc.com\\\/en\\\/#website\"},\"primaryImageOfPage\":{\"@id\":\"https:\\\/\\\/www.thinkupc.com\\\/en\\\/news\\\/ai-is-a-7-story-building\\\/#primaryimage\"},\"image\":{\"@id\":\"https:\\\/\\\/www.thinkupc.com\\\/en\\\/news\\\/ai-is-a-7-story-building\\\/#primaryimage\"},\"thumbnailUrl\":\"https:\\\/\\\/www.thinkupc.com\\\/wp-content\\\/uploads\\\/2026\\\/05\\\/edificiIAI-retall1-noticia.jpg\",\"datePublished\":\"2026-05-04T10:09:12+00:00\",\"dateModified\":\"2026-05-07T08:38:33+00:00\",\"description\":\"Javier Otero reflects on the challenges of agentic AI and the structure companies need to lead the digital revolution.\",\"breadcrumb\":{\"@id\":\"https:\\\/\\\/www.thinkupc.com\\\/en\\\/news\\\/ai-is-a-7-story-building\\\/#breadcrumb\"},\"inLanguage\":\"en-US\",\"potentialAction\":[{\"@type\":\"ReadAction\",\"target\":[\"https:\\\/\\\/www.thinkupc.com\\\/en\\\/news\\\/ai-is-a-7-story-building\\\/\"]}]},{\"@type\":\"ImageObject\",\"inLanguage\":\"en-US\",\"@id\":\"https:\\\/\\\/www.thinkupc.com\\\/en\\\/news\\\/ai-is-a-7-story-building\\\/#primaryimage\",\"url\":\"https:\\\/\\\/www.thinkupc.com\\\/wp-content\\\/uploads\\\/2026\\\/05\\\/edificiIAI-retall1-noticia.jpg\",\"contentUrl\":\"https:\\\/\\\/www.thinkupc.com\\\/wp-content\\\/uploads\\\/2026\\\/05\\\/edificiIAI-retall1-noticia.jpg\",\"width\":1400,\"height\":683,\"caption\":\"edificiIAI\"},{\"@type\":\"BreadcrumbList\",\"@id\":\"https:\\\/\\\/www.thinkupc.com\\\/en\\\/news\\\/ai-is-a-7-story-building\\\/#breadcrumb\",\"itemListElement\":[{\"@type\":\"ListItem\",\"position\":1,\"name\":\"Home\",\"item\":\"https:\\\/\\\/www.thinkupc.com\\\/en\\\/\"},{\"@type\":\"ListItem\",\"position\":2,\"name\":\"News\",\"item\":\"https:\\\/\\\/www.thinkupc.com\\\/en\\\/.\\\/news\\\/\"},{\"@type\":\"ListItem\",\"position\":3,\"name\":\"AI Is a 7-Story Building\"}]},{\"@type\":\"WebSite\",\"@id\":\"https:\\\/\\\/www.thinkupc.com\\\/en\\\/#website\",\"url\":\"https:\\\/\\\/www.thinkupc.com\\\/en\\\/\",\"name\":\"ThinkUPC\",\"description\":\"AI-driven Digital Transformation\",\"publisher\":{\"@id\":\"https:\\\/\\\/www.thinkupc.com\\\/en\\\/#organization\"},\"potentialAction\":[{\"@type\":\"SearchAction\",\"target\":{\"@type\":\"EntryPoint\",\"urlTemplate\":\"https:\\\/\\\/www.thinkupc.com\\\/en\\\/?s={search_term_string}\"},\"query-input\":{\"@type\":\"PropertyValueSpecification\",\"valueRequired\":true,\"valueName\":\"search_term_string\"}}],\"inLanguage\":\"en-US\"},{\"@type\":\"Organization\",\"@id\":\"https:\\\/\\\/www.thinkupc.com\\\/en\\\/#organization\",\"name\":\"ThinkUPC\",\"url\":\"https:\\\/\\\/www.thinkupc.com\\\/en\\\/\",\"logo\":{\"@type\":\"ImageObject\",\"inLanguage\":\"en-US\",\"@id\":\"https:\\\/\\\/www.thinkupc.com\\\/en\\\/#\\\/schema\\\/logo\\\/image\\\/\",\"url\":\"https:\\\/\\\/www.thinkupc.com\\\/wp-content\\\/uploads\\\/2025\\\/03\\\/logo-claim_B.png\",\"contentUrl\":\"https:\\\/\\\/www.thinkupc.com\\\/wp-content\\\/uploads\\\/2025\\\/03\\\/logo-claim_B.png\",\"width\":2325,\"height\":1027,\"caption\":\"ThinkUPC\"},\"image\":{\"@id\":\"https:\\\/\\\/www.thinkupc.com\\\/en\\\/#\\\/schema\\\/logo\\\/image\\\/\"},\"sameAs\":[\"https:\\\/\\\/bsky.app\\\/profile\\\/thinkupc.com\",\"https:\\\/\\\/es.linkedin.com\\\/company\\\/thinkupc\"]},{\"@type\":\"Person\",\"@id\":\"https:\\\/\\\/www.thinkupc.com\\\/en\\\/#\\\/schema\\\/person\\\/e9c4eb2ebe12f20d120c4e4513e613c7\",\"name\":\"Javier Otero\",\"image\":{\"@type\":\"ImageObject\",\"inLanguage\":\"en-US\",\"@id\":\"https:\\\/\\\/secure.gravatar.com\\\/avatar\\\/4461fc6c334f269d377c235d0b5e5af481941f29dfe83bdd6ffa17f49b4061c7?s=96&d=mm&r=g\",\"url\":\"https:\\\/\\\/secure.gravatar.com\\\/avatar\\\/4461fc6c334f269d377c235d0b5e5af481941f29dfe83bdd6ffa17f49b4061c7?s=96&d=mm&r=g\",\"contentUrl\":\"https:\\\/\\\/secure.gravatar.com\\\/avatar\\\/4461fc6c334f269d377c235d0b5e5af481941f29dfe83bdd6ffa17f49b4061c7?s=96&d=mm&r=g\",\"caption\":\"Javier Otero\"},\"url\":\"https:\\\/\\\/www.thinkupc.com\\\/en\\\/author\\\/javier-otero\\\/\"}]}<\/script>\n<!-- \/ Yoast SEO plugin. -->","yoast_head_json":{"title":"AI Is a 7-Story Building - ThinkUPC","description":"Javier Otero reflects on the challenges of agentic AI and the structure companies need to lead the digital revolution.","robots":{"index":"index","follow":"follow","max-snippet":"max-snippet:-1","max-image-preview":"max-image-preview:large","max-video-preview":"max-video-preview:-1"},"canonical":"https:\/\/www.thinkupc.com\/en\/news\/ai-is-a-7-story-building\/","og_locale":"en_US","og_type":"article","og_title":"AI Is a 7-Story Building - ThinkUPC","og_description":"Javier Otero reflects on the challenges of agentic AI and the structure companies need to lead the digital revolution.","og_url":"https:\/\/www.thinkupc.com\/en\/news\/ai-is-a-7-story-building\/","og_site_name":"ThinkUPC","article_published_time":"2026-05-04T10:09:12+00:00","article_modified_time":"2026-05-07T08:38:33+00:00","og_image":[{"width":1400,"height":683,"url":"https:\/\/www.thinkupc.com\/wp-content\/uploads\/2026\/05\/edificiIAI-retall1-noticia.jpg","type":"image\/jpeg"}],"author":"Javier Otero","twitter_card":"summary_large_image","twitter_misc":{"Written by":"Javier Otero","Est. reading time":"14 minutes"},"schema":{"@context":"https:\/\/schema.org","@graph":[{"@type":"Article","@id":"https:\/\/www.thinkupc.com\/en\/news\/ai-is-a-7-story-building\/#article","isPartOf":{"@id":"https:\/\/www.thinkupc.com\/en\/news\/ai-is-a-7-story-building\/"},"author":{"name":"Javier Otero","@id":"https:\/\/www.thinkupc.com\/en\/#\/schema\/person\/e9c4eb2ebe12f20d120c4e4513e613c7"},"headline":"AI Is a 7-Story Building","datePublished":"2026-05-04T10:09:12+00:00","dateModified":"2026-05-07T08:38:33+00:00","mainEntityOfPage":{"@id":"https:\/\/www.thinkupc.com\/en\/news\/ai-is-a-7-story-building\/"},"wordCount":3028,"publisher":{"@id":"https:\/\/www.thinkupc.com\/en\/#organization"},"image":{"@id":"https:\/\/www.thinkupc.com\/en\/news\/ai-is-a-7-story-building\/#primaryimage"},"thumbnailUrl":"https:\/\/www.thinkupc.com\/wp-content\/uploads\/2026\/05\/edificiIAI-retall1-noticia.jpg","keywords":["Digital Strategy"],"articleSection":["Digital Strategy","News"],"inLanguage":"en-US"},{"@type":"WebPage","@id":"https:\/\/www.thinkupc.com\/en\/news\/ai-is-a-7-story-building\/","url":"https:\/\/www.thinkupc.com\/en\/news\/ai-is-a-7-story-building\/","name":"AI Is a 7-Story Building - ThinkUPC","isPartOf":{"@id":"https:\/\/www.thinkupc.com\/en\/#website"},"primaryImageOfPage":{"@id":"https:\/\/www.thinkupc.com\/en\/news\/ai-is-a-7-story-building\/#primaryimage"},"image":{"@id":"https:\/\/www.thinkupc.com\/en\/news\/ai-is-a-7-story-building\/#primaryimage"},"thumbnailUrl":"https:\/\/www.thinkupc.com\/wp-content\/uploads\/2026\/05\/edificiIAI-retall1-noticia.jpg","datePublished":"2026-05-04T10:09:12+00:00","dateModified":"2026-05-07T08:38:33+00:00","description":"Javier Otero reflects on the challenges of agentic AI and the structure companies need to lead the digital revolution.","breadcrumb":{"@id":"https:\/\/www.thinkupc.com\/en\/news\/ai-is-a-7-story-building\/#breadcrumb"},"inLanguage":"en-US","potentialAction":[{"@type":"ReadAction","target":["https:\/\/www.thinkupc.com\/en\/news\/ai-is-a-7-story-building\/"]}]},{"@type":"ImageObject","inLanguage":"en-US","@id":"https:\/\/www.thinkupc.com\/en\/news\/ai-is-a-7-story-building\/#primaryimage","url":"https:\/\/www.thinkupc.com\/wp-content\/uploads\/2026\/05\/edificiIAI-retall1-noticia.jpg","contentUrl":"https:\/\/www.thinkupc.com\/wp-content\/uploads\/2026\/05\/edificiIAI-retall1-noticia.jpg","width":1400,"height":683,"caption":"edificiIAI"},{"@type":"BreadcrumbList","@id":"https:\/\/www.thinkupc.com\/en\/news\/ai-is-a-7-story-building\/#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https:\/\/www.thinkupc.com\/en\/"},{"@type":"ListItem","position":2,"name":"News","item":"https:\/\/www.thinkupc.com\/en\/.\/news\/"},{"@type":"ListItem","position":3,"name":"AI Is a 7-Story Building"}]},{"@type":"WebSite","@id":"https:\/\/www.thinkupc.com\/en\/#website","url":"https:\/\/www.thinkupc.com\/en\/","name":"ThinkUPC","description":"AI-driven Digital Transformation","publisher":{"@id":"https:\/\/www.thinkupc.com\/en\/#organization"},"potentialAction":[{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https:\/\/www.thinkupc.com\/en\/?s={search_term_string}"},"query-input":{"@type":"PropertyValueSpecification","valueRequired":true,"valueName":"search_term_string"}}],"inLanguage":"en-US"},{"@type":"Organization","@id":"https:\/\/www.thinkupc.com\/en\/#organization","name":"ThinkUPC","url":"https:\/\/www.thinkupc.com\/en\/","logo":{"@type":"ImageObject","inLanguage":"en-US","@id":"https:\/\/www.thinkupc.com\/en\/#\/schema\/logo\/image\/","url":"https:\/\/www.thinkupc.com\/wp-content\/uploads\/2025\/03\/logo-claim_B.png","contentUrl":"https:\/\/www.thinkupc.com\/wp-content\/uploads\/2025\/03\/logo-claim_B.png","width":2325,"height":1027,"caption":"ThinkUPC"},"image":{"@id":"https:\/\/www.thinkupc.com\/en\/#\/schema\/logo\/image\/"},"sameAs":["https:\/\/bsky.app\/profile\/thinkupc.com","https:\/\/es.linkedin.com\/company\/thinkupc"]},{"@type":"Person","@id":"https:\/\/www.thinkupc.com\/en\/#\/schema\/person\/e9c4eb2ebe12f20d120c4e4513e613c7","name":"Javier Otero","image":{"@type":"ImageObject","inLanguage":"en-US","@id":"https:\/\/secure.gravatar.com\/avatar\/4461fc6c334f269d377c235d0b5e5af481941f29dfe83bdd6ffa17f49b4061c7?s=96&d=mm&r=g","url":"https:\/\/secure.gravatar.com\/avatar\/4461fc6c334f269d377c235d0b5e5af481941f29dfe83bdd6ffa17f49b4061c7?s=96&d=mm&r=g","contentUrl":"https:\/\/secure.gravatar.com\/avatar\/4461fc6c334f269d377c235d0b5e5af481941f29dfe83bdd6ffa17f49b4061c7?s=96&d=mm&r=g","caption":"Javier Otero"},"url":"https:\/\/www.thinkupc.com\/en\/author\/javier-otero\/"}]}},"_links":{"self":[{"href":"https:\/\/www.thinkupc.com\/en\/wp-json\/wp\/v2\/posts\/14645","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.thinkupc.com\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.thinkupc.com\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.thinkupc.com\/en\/wp-json\/wp\/v2\/users\/30"}],"replies":[{"embeddable":true,"href":"https:\/\/www.thinkupc.com\/en\/wp-json\/wp\/v2\/comments?post=14645"}],"version-history":[{"count":22,"href":"https:\/\/www.thinkupc.com\/en\/wp-json\/wp\/v2\/posts\/14645\/revisions"}],"predecessor-version":[{"id":14907,"href":"https:\/\/www.thinkupc.com\/en\/wp-json\/wp\/v2\/posts\/14645\/revisions\/14907"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.thinkupc.com\/en\/wp-json\/wp\/v2\/media\/14891"}],"wp:attachment":[{"href":"https:\/\/www.thinkupc.com\/en\/wp-json\/wp\/v2\/media?parent=14645"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.thinkupc.com\/en\/wp-json\/wp\/v2\/categories?post=14645"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.thinkupc.com\/en\/wp-json\/wp\/v2\/tags?post=14645"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}