Manuel Hernández Giuliani

  What three years of working with AI taught me about building software without losing control A few days ago, I was talking with Rafa, a colleague and a friend who has shared many battles with me. I was trying to explain what I had learned after more than three years of hands-on research and work with artificial intelligence: months of development, experiments, broken and repaired code, prompts that seemed brilliant but were not, late nights, mistakes, and repeated trips back to square one. As we talked, I shared that everything is distilled into the idea far simpler than any collection of prompts: Architecture and governance before engineering. At the beginning, I thought the challenge was to make AI do more: write more code, understand instructions better, or solve a larger task with less intervention. Today, I believe that was the wrong question. The real challenge is to ensure that AI makes sound decisions within clear boundaries, without replacing human authority or causing u...

  DeepSeek V4 Pro vs. Kimi K3: What Months of Real-World AI Agent Work Taught Me This article is based entirely on my personal experience using both models extensively through Pi.dev and my ARAYA framework. It is an independent research exercise, not a sponsored comparison, a reproduction of marketing claims, or an analysis based on technology news. This Is Not a Benchmark Most comparisons between artificial intelligence models begin with public benchmarks, release announcements, pricing tables, or carefully prepared demonstrations. My comparison began somewhere very different: inside real repositories, with incomplete context, conflicting documents, failing gates, distributed agent responsibilities, Git history , runtime evidence, and requirements that could not be considered complete merely because the generated code looked correct. Over several months, weeks, and many hours of intensive use, I worked with DeepSeek V4 Pro and Kimi K3 through Pi.dev while developing and operat...

 The AI industry may be entering a strange phase where narrative engineering advances faster than the underlying engineering itself. That does not mean frontier AI is fake. It is not. The progress is real. Models are improving quickly in coding, search, reasoning assistance, content generation, and cybersecurity analysis. Anyone working seriously in enterprise technology can already see the productivity impact. AI is becoming operationally useful. But something else is happening in parallel. Every few months, another company announces that its newest model is so powerful, so dangerous, or so transformative that humanity must proceed carefully. The language changes slightly each cycle, but the structure remains almost identical: existential concern, dramatic warnings, selective access, media amplification, institutional reaction, investor excitement, and another jump in valuation. At some point, it becomes reasonable to ask whether we are witnessing purely a technological revolution...

    This started while I was preparing my Big Data labs for my students. I wanted something closer to reality. Not only pipelines and Spark jobs, but how AI agents would actually interact with a governed data platform. So I set the environment as it should be in a serious setup. A lakehouse governed with AWS Lake Formation, metadata centralized in AWS Glue Data Catalog, and Spark handling execution. Clean, controlled, auditable. Then I added the missing piece. An agent that needs to understand intent and retrieve context, not just run queries. And almost immediately, the same request appeared. “We need a graph database for ontology.” In AWS terms, that means Amazon Neptune. I see this pattern often, not only with students. Also in real projects. Someone comes with a solution already decided. I always give the same answer. What is the business problem you are trying to solve? Because “I need Neptune” is not a requirement. It is a conclusion. When you force the conversation back...

Over the past few years, artificial intelligence has moved from promise to pervasive investment. Across industries, organizations have committed significant resources to AI initiatives. New models, new platforms, new capabilities. The pace of innovation has been extraordinary. And yet, a new question is emerging. Not what can we build with AI, but what is actually delivering measurable value. This shift marks an inflection point. We are entering a new phase of AI maturity. One defined not by experimentation, but by accountability. The Illusion of Progress From the outside, the landscape looks impressive. More models. More pilots. More demonstrations of capability. But inside many organizations, a different reality is taking shape. AI initiatives often remain disconnected from measurable business outcomes. Value is assumed, but not always quantified. Success is described in terms of activity, not impact. This is not a failure of technology. It is a reflection of something de...

  LinkedIn What we may be witnessing right now feels uncomfortably familiar. Not in the technology itself… but in the economics behind it. We are seeing hyperscalers committing extraordinary levels of capital, with projected AI-driven capex in 2026 approaching $650–700 billion annually, up from roughly $380–410 billion in 2025. In just one year, that is a 60–70%+ increase, with an estimated ~75% of that spend directly tied to AI infrastructure. Individually, the scale is even more striking: - Amazon alone is projected around $200B - Alphabet around $175–185B - Meta between $115–135B - Microsoft exceeding $100B+ run rate - Oracle adding another ~$50B This is no longer incremental investment. This is a structural capital shift at global scale. And this is where the parallel with the dot-com era becomes difficult to ignore. Back then, the belief was that the internet would transform everything. It did. But the timing of value creation and the scale of investment were completely misali...

  Original Post in LinkedIn I have been saying for a while that AI is no longer just about models… it is about systems that act. Now regulators are catching up. The Spanish Data Protection Agency (AEPD) has published guidance on agentic AI from a data protection perspective. And this is important. Because agentic AI is fundamentally different. These are not systems that just answer… they decide, interact, and execute autonomously to achieve goals. That changes everything. It changes how data is accessed, how decisions are made, and how responsibility is assigned. And one key message stands out: using AI does not remove accountability. Organizations remain fully responsible for data protection, transparency, and control. From an architecture perspective, this reinforces something we often underestimate. You cannot build agentic AI on top of weak data foundations. Governance, data minimization, traceability, and privacy-by-design are no longer compliance topics… they are cor...