---
title: 5sigmas
description: Artificial intelligence explained with rigor through visual explanations, learning series, primary sources and production engineering.
hide:
  - navigation
  - toc
  - footer
---

<div class="s5-landing">

<section class="s5-home-cover">
  <div class="s5-home-cover__intro">
    <div class="s5-eyebrow">5sigmas · Artificial intelligence, without the noise</div>
    <h1>See AI.<br />Understand how it actually works.</h1>
    <p>Each topic connects a visual explanation, an interactive mechanism, technical depth and primary sources. Build intuition first, manipulate the mechanism, inspect its limits and then go to the evidence.</p>

    <div class="s5-learning-loop" role="list" aria-label="5sigmas learning method">
      <div class="s5-learning-loop__item" role="listitem">
        <span class="s5-learning-loop__n">01</span>
        <span><strong>See</strong><small>Orient yourself around one idea</small></span>
      </div>
      <div class="s5-learning-loop__item" role="listitem">
        <span class="s5-learning-loop__n">02</span>
        <span><strong>Manipulate</strong><small>Change variables and inspect the mechanism</small></span>
      </div>
      <div class="s5-learning-loop__item" role="listitem">
        <span class="s5-learning-loop__n">03</span>
        <span><strong>Go deeper</strong><small>Technical chapter and primary sources</small></span>
      </div>
    </div>
  </div>

  <article class="s5-start-card">
    <header class="s5-start-card__header">
      <span>Recommended path</span>
      <span>01 · Reasoning Models</span>
    </header>

    <a class="s5-start-card__media" href="/en/series/modelos-razonadores/01-que-es-razonar/" aria-label="Open the introduction to reasoning models">
      <img src="/en/series/modelos-razonadores/01-que-es-razonar.jpg" alt="Native English visual explaining what reasoning models optimize during training and inference" />
    </a>

    <a class="s5-start-card__cta" href="/en/series/modelos-razonadores/01-que-es-razonar/">
      <span>
        <small>Start with one mechanism</small>
        <strong>What does it mean for a model to reason?</strong>
      </span>
      <span class="s5-start-card__cta-meta">Chapter · 9 min</span>
    </a>

    <div class="s5-start-card__detail">
      <p>Separate fluent generation from deliberate inference, then connect reasoning traces, reinforcement signals and test-time compute to what the system can actually verify.</p>
      <a class="s5-start-card__chapter" href="/en/series/modelos-razonadores/00_presentacion_serie/">
        <span>Open the complete series</span>
        <small>5 chapters →</small>
      </a>
    </div>
  </article>
</section>

<section class="s5-why" aria-labelledby="s5-why-title">
  <div class="s5-why__copy">
    <div class="s5-eyebrow">The standard behind the name</div>
    <h2 id="s5-why-title">Why 5sigmas?</h2>
    <p>In particle physics, a result is conventionally treated as a discovery only when it reaches <strong>five-sigma significance</strong>: a threshold stringent enough that an extreme background fluctuation has a probability of roughly one in 3.5 million under the usual Gaussian interpretation.</p>
    <p>5sigmas applies the same editorial instinct to artificial intelligence: separate durable knowledge from daily noise, demand evidence, and build judgment before accepting a narrative, tool or promise.</p>
    <div class="s5-why__metrics" aria-label="Meaning of five sigma">
      <span><strong>5σ</strong><small>discovery threshold</small></span>
      <span><strong>≈ 1 in 3.5M</strong><small>extreme-noise probability</small></span>
      <span><strong>Signal &gt; noise</strong><small>editorial principle</small></span>
    </div>
  </div>
  <figure class="s5-why__visual">
    <img src="/en/assets/images/design/five-sigma-threshold.svg" alt="Normal distribution with the five-sigma threshold separated from expected noise" />
    <figcaption>Do not publish because something is new. Publish when the explanation survives evidence, context and its own limitations.</figcaption>
  </figure>
</section>

<section class="s5-entry-section" aria-labelledby="s5-entry-title">
  <div class="s5-section-head s5-section-head--compact">
    <h2 id="s5-entry-title">Choose how to continue</h2>
  </div>
  <nav class="s5-entry-grid" aria-label="Ways to enter the English 5sigmas library">
    <a class="s5-entry" href="/en/temas/">
      <div class="s5-entry__index">01</div>
      <div class="s5-entry__title">Understand</div>
      <div class="s5-entry__text">Answer-first guides to LLMs, Transformers, reasoning, evaluation, agents and security.</div>
    </a>
    <a class="s5-entry" href="/en/series/">
      <div class="s5-entry__index">02</div>
      <div class="s5-entry__title">Learn</div>
      <div class="s5-entry__text">Ordered learning paths that build a technical mental model progressively.</div>
    </a>
    <a class="s5-entry" href="/en/articulos-tecnicos/">
      <div class="s5-entry__index">03</div>
      <div class="s5-entry__title">Build</div>
      <div class="s5-entry__text">Architecture, trade-offs and failure boundaries from real AI systems.</div>
    </a>
  </nav>
</section>

<section class="s5-section" aria-labelledby="s5-concepts-title-en">
  <div class="s5-section-head">
    <div>
      <h2 id="s5-concepts-title-en">Core concepts</h2>
      <p class="s5-section-head__copy">Direct answers to the concepts behind modern AI systems.</p>
    </div>
    <a href="/en/temas/">View all concepts →</a>
  </div>
  <div class="s5-simple-list">
    <a class="s5-list-row" href="/en/temas/llms/">
      <span class="s5-list-row__n">01</span>
      <span class="s5-list-row__title">LLMs</span>
      <span class="s5-list-row__desc">What large language models are, how they learn and how they generate tokens.</span>
      <span class="s5-list-row__meta">Models</span>
    </a>
    <a class="s5-list-row" href="/en/temas/transformer/">
      <span class="s5-list-row__n">02</span>
      <span class="s5-list-row__title">Transformers</span>
      <span class="s5-list-row__desc">Attention, representations and the architecture behind modern language models.</span>
      <span class="s5-list-row__meta">Architecture</span>
    </a>
    <a class="s5-list-row" href="/en/temas/razonamiento/">
      <span class="s5-list-row__n">03</span>
      <span class="s5-list-row__title">Reasoning</span>
      <span class="s5-list-row__desc">What reasoning means for an LLM and what changes at inference time.</span>
      <span class="s5-list-row__meta">Inference</span>
    </a>
    <a class="s5-list-row" href="/en/temas/evaluacion-modelos/">
      <span class="s5-list-row__n">04</span>
      <span class="s5-list-row__title">Model evaluation</span>
      <span class="s5-list-row__desc">How to measure real capability without confusing a benchmark with the objective.</span>
      <span class="s5-list-row__meta">Evaluation</span>
    </a>
    <a class="s5-list-row" href="/en/temas/agentes-ia/">
      <span class="s5-list-row__n">05</span>
      <span class="s5-list-row__title">AI agents</span>
      <span class="s5-list-row__desc">Goals, tools, state and boundaries that turn model responses into actions.</span>
      <span class="s5-list-row__meta">Systems</span>
    </a>
    <a class="s5-list-row" href="/en/temas/prompt-injection/">
      <span class="s5-list-row__n">06</span>
      <span class="s5-list-row__title">Prompt injection</span>
      <span class="s5-list-row__desc">Why untrusted data can influence instructions and which controls reduce the risk.</span>
      <span class="s5-list-row__meta">Security</span>
    </a>
  </div>
</section>

<section class="s5-section">
  <div class="s5-section-head">
    <h2>Visual explanations</h2>
    <a href="/en/series/modelos-razonadores/00_presentacion_serie/">Open the complete path →</a>
  </div>
  <div class="s5-media-stack">
    <a class="s5-media-card" href="/en/series/modelos-razonadores/02-fallos/">
      <img class="s5-media-card__image" src="/en/series/modelos-razonadores/02-fallos.jpg" alt="Native English visual about reasoning-model failure modes" />
      <div class="s5-media-card__meta">Visual · Reasoning</div>
      <div class="s5-media-card__title">Why a convincing reasoning trace can still fail</div>
    </a>
    <a class="s5-media-card" href="/en/series/modelos-razonadores/04-latencia-streaming/">
      <img class="s5-media-card__image" src="/en/series/modelos-razonadores/04-latencia-streaming.jpg" alt="Native English visual about latency and streaming in reasoning systems" />
      <div class="s5-media-card__meta">Visual · Interaction</div>
      <div class="s5-media-card__title">Latency is part of model quality once a human is waiting</div>
    </a>
    <a class="s5-media-card" href="/en/series/modelos-razonadores/05-riesgos/">
      <img class="s5-media-card__image" src="/en/series/modelos-razonadores/05-riesgos.jpg" alt="Native English visual about risks from additional inference-time reasoning" />
      <div class="s5-media-card__meta">Visual · Reliability</div>
      <div class="s5-media-card__title">More inference compute is a control knob, not a guarantee</div>
    </a>
  </div>
</section>

<section class="s5-section">
  <div class="s5-section-head">
    <h2>Choose a path</h2>
    <a href="/en/series/">View all series →</a>
  </div>
  <div class="s5-simple-list">
    <a class="s5-list-row" href="/en/series/fundamentos-ia-iag/00_presentacion_serie/">
      <span class="s5-list-row__n">01</span>
      <span class="s5-list-row__title">AI and Generative AI Foundations</span>
      <span class="s5-list-row__desc">The essential distinctions between software, learning, generation and AGI.</span>
      <span class="s5-list-row__meta">4 chapters</span>
    </a>
    <a class="s5-list-row" href="/en/series/multimodalidad-iag/00_presentacion_serie/">
      <span class="s5-list-row__n">02</span>
      <span class="s5-list-row__title">Multimodality</span>
      <span class="s5-list-row__desc">Text, image, audio and video inside the same generative system.</span>
      <span class="s5-list-row__meta">5 chapters</span>
    </a>
    <a class="s5-list-row" href="/en/series/modelos-razonadores/00_presentacion_serie/">
      <span class="s5-list-row__n">03</span>
      <span class="s5-list-row__title">Reasoning Models</span>
      <span class="s5-list-row__desc">Capability, failure, cost and latency when models spend compute before answering.</span>
      <span class="s5-list-row__meta">5 chapters</span>
    </a>
    <a class="s5-list-row" href="/en/series/datacenters-espacio/00_presentacion_serie/">
      <span class="s5-list-row__n">04</span>
      <span class="s5-list-row__title">Data Centers in Space</span>
      <span class="s5-list-row__desc">The physics and infrastructure behind orbital compute.</span>
      <span class="s5-list-row__meta">4 chapters</span>
    </a>
  </div>
</section>

<section class="s5-section">
  <div class="s5-section-head"><h2>Engineering</h2></div>
  <div class="s5-note-feature">
    <div>
      <div class="s5-eyebrow">Technical note</div>
      <h2>Proactive and reactive agents and tool calls</h2>
      <p>How to separate the conversational contract, internal execution and deferred completion without turning conversation history into the system database.</p>
      <a class="s5-text-link" href="/en/articulos-tecnicos/proactive-reactive-agent-and-tool-calls/">Open the note →</a>
    </div>
    <div class="s5-note-feature__meta">Conversational runtime<br />Internal state<br />Asynchronous execution</div>
  </div>
</section>

</div>