# Santiago Santa María Morales — EUREKA

> Ideas, talks and analysis on applied AI, agentic AI, robotics, business and technology to improve organizations and lives.

Author: Santiago Santa María Morales — Head of AI Forward Deployed Engineers, LATAM (Google Cloud); physicist by training.
Languages: English (/en), Español (/es), Português (/pt). Default: English.

## Key pages

- [Agentic AI Knowledge Base](https://santismm.com/en/knowledge): A reference knowledge base on agentic AI, enterprise AI, harness engineering, AI architecture and AI governance — self-contained, citable units for humans and AI agents.
- [Enterprise AI Patterns Library](https://santismm.com/en/patterns): A public library of reusable enterprise AI and agentic design patterns — problem, context, solution, benefits, risks and when not to use — for humans and AI agents.
- [Harness Engineering](https://santismm.com/en/harness-engineering): The reference hub for harness engineering — designing the tools, memory, context, control loops and evaluation that turn model capability into reliable agentic systems.
- [About](https://santismm.com/en/about): About Santiago Santa María Morales — Head of AI FDE LATAM at Google Cloud, ex-NTT DATA Syntphony, physicist by training.
- [Library](https://santismm.com/en/library): Everything I've published — articles, videos and podcasts — across Substack, Medium, YouTube and LinkedIn. Filter by kind or topic.
- [Labs](https://santismm.com/en/labs): SANTISMM Labs — experiments, benchmarks, simulators and interactive tools to explore AI, agents, robotics and complex systems.
- [Talks](https://santismm.com/en/talks): Keynotes, panels and workshops on agentic AI, foundation models and conversational AI applied to business and science.
- [For AI agents](https://santismm.com/en/for-agents): Every machine-readable surface of santismm.com in one place: the MCP endpoint, llms.txt, llms-full.txt, ai-index.json, the JSON APIs, the knowledge graph and the Agent Skill.

## Harness Engineering Handbook

The canonical long-form corpus on harness engineering (HRN-001…014): definition, history, taxonomy, principles, memory, observability, evaluation, governance, planning, orchestration, security, case studies, glossary and bibliography. Each chapter also has a JSON API at /api/handbook/{id}.

- [HRN-001 · Harness Engineering: Definition and Overview](https://santismm.com/en/handbook/definition-and-overview): Harness Engineering is the discipline of building reliable agentic systems for enterprise environments — the engineered scaffolding of memory, tools, orchestration, observability, evaluation, governance, and security that surrounds the model. (evidence: theoretical, confidence: medium)
- [HRN-002 · A Brief History of Harness Engineering](https://santismm.com/en/handbook/a-brief-history-of-harness-engineering): How the field moved from prompt engineering to tool use to agents to harnesses, and why the engineered scaffolding around the model became its own discipline. (evidence: theoretical, confidence: medium)
- [HRN-003 · The Harness Taxonomy](https://santismm.com/en/handbook/the-harness-taxonomy): A structured taxonomy of the harness — memory, tools, planning, orchestration, observability, evaluation, governance, and security — naming each component, its responsibility, and how the parts compose into a reliable agentic system. (evidence: theoretical, confidence: medium)
- [HRN-004 · Harness Engineering Principles](https://santismm.com/en/handbook/harness-engineering-principles): The core engineering principles of the harness — reliability over capability, determinism boundaries, observability-first, evidence-first, defense in depth, least authority, graceful degradation, and idempotent actuation — that hold across every component. (evidence: theoretical, confidence: medium)
- [HRN-005 · Memory in Agentic Systems](https://santismm.com/en/handbook/memory-in-agentic-systems): How the harness governs what the model sees and remembers — working, short-term, and long-term memory; the context window as a budget; retrieval, compression, and deliberate forgetting. (evidence: theoretical, confidence: medium)
- [HRN-006 · Observability for Agentic Systems](https://santismm.com/en/handbook/observability-for-agentic-systems): How to make a non-deterministic, multi-step agent inspectable — traces and spans, token and cost accounting, evaluation hooks, and deterministic replay — so the system can be debugged, measured, and trusted. (evidence: theoretical, confidence: medium)
- [HRN-007 · Evaluation of Agentic Systems](https://santismm.com/en/handbook/evaluation-of-agentic-systems): How to measure whether an agent is actually good and getting better — offline and online evaluation, golden sets, LLM-as-judge, regression suites, and task-completion metrics — turning agent development from craft into engineering. (evidence: theoretical, confidence: medium)
- [HRN-008 · Governance within the Harness](https://santismm.com/en/handbook/governance-within-the-harness): Governance is an engineered harness layer that enforces policy, approvals, and guardrails at runtime, turning enterprise AI obligations into executable controls that gate every agent action. (evidence: theoretical, confidence: medium)
- [HRN-009 · Planning and Goal Management](https://santismm.com/en/handbook/planning-and-goal-management): Planning and goal management is the harness layer that decomposes goals into executable plans, represents and tracks plan state, and replans under failure, making agent autonomy directed rather than reactive. (evidence: theoretical, confidence: medium)
- [HRN-010 · Orchestration](https://santismm.com/en/handbook/orchestration): Orchestration is the harness layer that drives execution—single vs multi-agent topologies, supervisor/worker delegation, routing, state machines, and durable workflows—turning a plan into reliable, resumable action. (evidence: theoretical, confidence: medium)
- [HRN-011 · Security for Agentic Systems](https://santismm.com/en/handbook/security-for-agentic-systems): Security for agentic systems is the harness layer that defends against prompt injection, sandboxes tools and permissions, prevents data exfiltration, and enforces agent identity and least privilege across every action. (evidence: theoretical, confidence: medium)
- [HRN-012 · Case Studies in Harness Engineering](https://santismm.com/en/handbook/case-studies-in-harness-engineering): Three representative, anonymized composite case studies showing how harness layers—memory, planning, orchestration, governance, security, observability—combine end-to-end to make enterprise agents reliable. (evidence: theoretical, confidence: medium)
- [HRN-013 · Glossary](https://santismm.com/en/handbook/glossary): A canonical glossary of Harness Engineering and agentic-systems terminology—harness, agent, tool, orchestration, evaluation, span, RAG, MCP, guardrail, and more—with crisp, citable definitions. (evidence: theoretical, confidence: high)
- [HRN-014 · Bibliography](https://santismm.com/en/handbook/bibliography): A curated, themed reading list for Harness Engineering—agents and orchestration, evaluation, security, governance and standards, and protocols—covering foundational papers, industry writeups, and regulatory frameworks. (evidence: theoretical, confidence: high)

## Knowledge Base

Self-contained, citable units on agentic and enterprise AI. Each unit also has a JSON API at /api/knowledge/{slug}.

- [What are Agent Memory Systems?](https://santismm.com/en/knowledge/agent-memory): Agent memory is the set of mechanisms an AI agent uses to store, organize and retrieve information beyond its immediate context window, spanning short-term working memory and long-term persistent memory. (evidence: industry_observation, confidence: high)
- [What is Agentic AI?](https://santismm.com/en/knowledge/agentic-ai): Agentic AI is the class of AI systems that autonomously plan and execute multi-step tasks by combining a model with memory, tools and a control loop. (evidence: industry_observation, confidence: high)
- [What is Agentic AI Evaluation?](https://santismm.com/en/knowledge/agentic-evaluation): Agentic evaluation is the measurement of an AI agent's end-to-end task performance — success rate, reliability, cost and safety — on realistic, multi-step tasks in an environment. (evidence: production, confidence: medium)
- [What is an AI Agent?](https://santismm.com/en/knowledge/ai-agent): An AI agent is a system that pairs a reasoning model with tools, memory and an orchestration loop so it can plan and act over multiple steps to achieve a goal. (evidence: industry_observation, confidence: high)
- [What is AI Governance?](https://santismm.com/en/knowledge/ai-governance): AI governance is the framework of policies, processes, roles and controls an organization uses to manage AI risk and ensure AI systems are responsible, compliant, secure and accountable across their lifecycle. (evidence: industry_observation, confidence: high)
- [What is AI Agent Observability?](https://santismm.com/en/knowledge/ai-observability): AI observability is the practice of capturing traces, metrics and logs of an AI system's behavior — every prompt, tool call, retrieval, output, token, latency and cost — so its decisions can be understood, debugged and improved. (evidence: production, confidence: medium)
- [What is Context Engineering?](https://santismm.com/en/knowledge/context-engineering): Context engineering is the practice of curating, compressing and sequencing the information placed in a model's context window so it has the most relevant signal — and the least noise — at each step. (evidence: industry_observation, confidence: high)
- [What are Embeddings & Vector Search?](https://santismm.com/en/knowledge/embeddings): An embedding is a dense numeric vector that encodes the meaning of a piece of data, positioned so that similar items are close in vector space; vector search retrieves the nearest embeddings to a query. (evidence: benchmark, confidence: high)
- [What is Enterprise RAG?](https://santismm.com/en/knowledge/enterprise-rag): Enterprise RAG is a pattern that retrieves relevant passages from an organization's governed knowledge sources and supplies them to a model as context, so answers are grounded, current and citable. (evidence: benchmark, confidence: high)
- [What is Fine-tuning?](https://santismm.com/en/knowledge/fine-tuning): Fine-tuning is the process of further training a pretrained model on a focused dataset to adapt its weights toward a specific behavior, style, format or domain. (evidence: benchmark, confidence: high)
- [What are Foundation Models?](https://santismm.com/en/knowledge/foundation-models): A foundation model is a large, general-purpose model pretrained on broad data that serves as a base which can be adapted — via prompting, retrieval or fine-tuning — to many downstream tasks. (evidence: benchmark, confidence: high)
- [What are AI Guardrails?](https://santismm.com/en/knowledge/guardrails): AI guardrails are runtime safeguards that validate, filter or constrain a model's inputs, outputs and actions to keep its behavior safe, compliant and within defined policy. (evidence: industry_observation, confidence: high)
- [What is Harness Engineering?](https://santismm.com/en/knowledge/harness-engineering): Harness engineering is the practice of designing, building and optimizing the scaffolding (tools, memory, prompts, environment and control loop) that turns a model's raw capability into reliable, goal-directed action. (evidence: theoretical, confidence: medium)
- [What is the Human-in-the-Loop Pattern?](https://santismm.com/en/knowledge/human-in-the-loop): Human-in-the-loop is a pattern in which a human reviews, approves, edits or rejects an AI system's proposed output or action before it is executed, inserting human judgment at defined decision points. (evidence: industry_observation, confidence: high)
- [What is the Model Context Protocol (MCP)?](https://santismm.com/en/knowledge/model-context-protocol): MCP is an open protocol that standardizes how AI applications provide tools, resources and context to models through a common client–server interface. (evidence: industry_observation, confidence: high)
- [What is a Multi-Agent Architecture?](https://santismm.com/en/knowledge/multi-agent-architecture): A multi-agent architecture is a system design in which multiple specialized AI agents coordinate — through an orchestrator, a pipeline or peer interaction — to accomplish a task that is decomposed across them. (evidence: industry_observation, confidence: high)
- [What is Prompt Engineering?](https://santismm.com/en/knowledge/prompt-engineering): Prompt engineering is the practice of designing and refining the instructions, context and examples given to a language model to reliably elicit a desired output. (evidence: benchmark, confidence: high)
- [What is Prompt Injection?](https://santismm.com/en/knowledge/prompt-injection): Prompt injection is a security attack where adversarial instructions embedded in untrusted input cause a language model to deviate from its intended behavior, bypass safeguards, or perform unintended actions. (evidence: industry_observation, confidence: high)
- [What are Reasoning Models?](https://santismm.com/en/knowledge/reasoning-models): Reasoning models are language models optimized to perform extended step-by-step reasoning at inference time — using additional test-time compute — to improve accuracy on complex, multi-step problems. (evidence: benchmark, confidence: high)
- [What is Tool Use (Function Calling)?](https://santismm.com/en/knowledge/tool-use): Tool use (function calling) is the capability that lets a language model call predefined external functions or APIs, with arguments it generates, and incorporate the results into its response or next step. (evidence: production, confidence: medium)

## Enterprise AI Patterns Library

Reusable design patterns for AI and agentic systems. Each pattern also has a JSON API at /api/patterns/{slug}.

- [Context Compression](https://santismm.com/en/patterns/context-compression): Context compression reduces the tokens fed to a model on each call while preserving the information it actually needs to act. Use it on long-running agents and long conversations to cut cost and latency and to stay inside the context window. The three levers are summarizing history, pruning irrelevant context, and compressing prompts. The central risk is lossy: dropping the one detail that mattered. Measure information retained, not just tokens saved. (evidence: production, confidence: low)
- [Evaluator-Optimizer](https://santismm.com/en/patterns/evaluator-optimizer): One LLM generates a response while a second LLM evaluates it against criteria and returns feedback; the generator revises and the loop repeats until the evaluation passes. It raises quality on tasks with clear evaluation criteria, at the cost of extra calls. (evidence: industry_observation, confidence: high)
- [Goal Decomposition](https://santismm.com/en/patterns/goal-decomposition): Goal decomposition has an agent break a high-level goal into an ordered set of smaller, tractable sub-tasks — a plan — before acting, then execute and monitor that plan, re-planning when steps fail. The explicit plan becomes an inspectable artifact you can review, gate, and debug. Use it when a goal needs several dependent steps and reactive, step-at-a-time agents drift or stall; skip it for simple, single-shot tasks. (evidence: industry_observation, confidence: high)
- [Human Approval Gate](https://santismm.com/en/patterns/human-approval-gate): A human approval gate pauses an automated workflow at a defined checkpoint so a person can review, edit or reject a proposed action before it executes — especially for high-impact, irreversible or regulated operations. It is the operational form of human-in-the-loop oversight. (evidence: industry_observation, confidence: high)
- [Human Escalation](https://santismm.com/en/patterns/human-escalation): Hand the whole task to a human when the agent detects it is out of its depth — low confidence, repeated failure, ambiguity, or sensitive situations — and pass full context so the person can take over without re-investigating. Unlike an approval gate, which pauses one action for sign-off, escalation transfers ownership so the agent stops driving. The hard part is calibrating triggers to avoid both over- and under-escalation. (evidence: industry_observation, confidence: high)
- [Long-Term Memory](https://santismm.com/en/patterns/long-term-memory): Give an agent persistent memory across sessions so it remembers facts, user preferences, and prior outcomes beyond a single context window. A write path decides what to store, summarizes it, and deduplicates it; a read path retrieves only the relevant memories into context when needed. Unlike semantic caching, which caches whole answers to skip recomputation, long-term memory stores durable facts and state and recomposes them into fresh reasoning each time. (evidence: production, confidence: low)
- [Orchestrator-Workers](https://santismm.com/en/patterns/orchestrator-workers): An orchestrator LLM dynamically breaks a task into subtasks, delegates each to a worker LLM, and synthesizes the results. Unlike fixed parallelization, the orchestrator decides the subtasks at runtime — making it suited to complex tasks whose decomposition is not known in advance. (evidence: production, confidence: low)
- [Parallelization](https://santismm.com/en/patterns/parallelization): Parallelization runs multiple LLM calls at the same time and aggregates the results. Two flavors: sectioning (split a task into independent subtasks run in parallel) and voting (run the same task several times to improve reliability or coverage). It cuts latency and can raise quality. (evidence: industry_observation, confidence: high)
- [Prompt Chaining](https://santismm.com/en/patterns/prompt-chaining): Prompt chaining decomposes a task into a fixed sequence of LLM calls, where each step works on the output of the previous one. It trades a little latency for much higher accuracy and control, and is the simplest workflow pattern: use it whenever a task cleanly splits into ordered subtasks. (evidence: industry_observation, confidence: high)
- [Recovery Strategy](https://santismm.com/en/patterns/recovery-strategy): Give the agent an explicit plan for when things break. Detect failures by validating outputs and catching tool errors; then retry with adjustment, fall back to an alternative path, roll back partial actions, or escalate. Bound retries to avoid runaway loops and cost, make actions idempotent, and distinguish transient from permanent failures. The goal is graceful degradation instead of crashes or silently wrong results. (evidence: production, confidence: low)
- [Reflection](https://santismm.com/en/patterns/reflection): Reflection has a model critique its own output and then revise it, using the critique as feedback. It is a lightweight, single-model way to catch mistakes and improve quality on reasoning, coding and writing tasks — at the cost of extra calls. (evidence: industry_observation, confidence: high)
- [Routing](https://santismm.com/en/patterns/routing): Routing classifies an input and directs it to the most appropriate specialized handler, prompt or model. It improves quality by letting each path be optimized for its case, and controls cost by sending easy requests to cheap models and hard ones to capable models. (evidence: production, confidence: low)
- [Semantic Caching](https://santismm.com/en/patterns/semantic-caching): Semantic caching stores past model responses and reuses them when a new request is semantically similar to a previous one — matching by meaning via embeddings, not exact text. It cuts cost and latency for repetitive or near-duplicate queries common in production. (evidence: production, confidence: low)
- [Supervisor Agent](https://santismm.com/en/patterns/supervisor-agent): A supervisor agent is a persistent coordinator that manages a team of specialized sub-agents. It reads the conversation state, decides which specialist should act next, routes messages to it, and integrates returned results toward the goal. Unlike a one-shot decomposer, the supervisor stays in the loop across many turns, delegating by capability and re-planning until the task is done or handed back to the user. (evidence: industry_observation, confidence: high)
- [Task Prioritization](https://santismm.com/en/patterns/task-prioritization): Order an agent's candidate tasks by value, urgency, dependencies, and cost instead of processing them first-in-first-out. A scoring function and a priority queue decide what runs next, so limited compute, budget, and time go to the work that matters most. Re-score as state changes, and bound the queue so it cannot grow without limit. (evidence: production, confidence: low)

## Enterprise AI Reference Architectures

End-to-end blueprints composing patterns and knowledge into working enterprise systems. Each architecture also has a JSON API at /api/architectures/{slug}.

- [ARCH-001 · Customer Service Agent](https://santismm.com/en/architectures/customer-service-agent): A reference architecture for an enterprise customer-service agent that resolves common requests end to end — answering from a grounded knowledge base, acting in the CRM and ticketing systems through tools, and escalating to a human when confidence is low or the action is high-impact. It pairs retrieval for grounding with risk-based human approval for safety, and is observable so every conversation can be evaluated and improved. (evidence: industry_observation, confidence: medium)
- [ARCH-002 · Enterprise Knowledge Assistant](https://santismm.com/en/architectures/enterprise-knowledge-assistant): A reference architecture for an internal knowledge assistant that answers employee questions from the company's own documents — wikis, policies, tickets, code — with citations and respecting each user's access permissions. It combines hybrid retrieval and reranking for grounding, permission-aware filtering for security, and an evaluation harness so answer quality is measured rather than assumed. The hard parts are not the model; they are retrieval quality, access control and evaluation. (evidence: industry_observation, confidence: high)
- [ARCH-003 · Sales Copilot](https://santismm.com/en/architectures/sales-copilot): A sales copilot is an agent that assists reps end to end: it researches accounts from CRM and product data, drafts personalized outreach and follow-ups, logs activity, surfaces next-best-actions, and prepares meeting briefs. Everything is grounded in the company's CRM, deal, and product knowledge to avoid hallucinated claims about features or pricing. The rep stays in control: a human-approval gate sits in front of any outbound action, so the agent drafts but never sends to a customer on its own. Success is measured honestly through rep productivity and pipeline impact, not vanity activity counts. (evidence: industry_observation, confidence: medium)
- [ARCH-004 · AI Workforce](https://santismm.com/en/architectures/ai-workforce): An AI workforce is an orchestrated team of specialized agents that collaborate on multi-step business processes under a supervisor. The supervisor decomposes a goal, prioritizes and delegates subtasks to specialist agents (research, drafting, QA), and they share state through a common store. Agents escalate to humans when out of depth, and every step is observable and governed. Treat it as managing digital workers: success depends less on any single agent and more on coordination, clear ownership via an agent registry, human oversight, and evaluating the whole team rather than parts. (evidence: industry_observation, confidence: medium)
- [ARCH-005 · Operations Center](https://santismm.com/en/architectures/operations-center): An operations center is an agentic AIOps system that watches monitoring signals and alerts, correlates and triages them, diagnoses likely root cause, and executes only vetted runbook remediations — keeping destructive or novel actions behind human approval. It cuts alert fatigue and shortens mean time to resolution by automating safe, read-mostly diagnostics while escalating risky writes to on-call engineers. Every action is audited and reversible. Success is measured honestly with MTTR, false-action rate, and escalation precision, not by automation volume. (evidence: industry_observation, confidence: medium)

## Enterprise AI Governance

Practical reference units for governing enterprise AI — major regimes (EU AI Act, ISO/IEC 42001, NIST AI RMF) and an agentic-AI governance checklist. Each unit also has a JSON API at /api/governance/{slug}.

- [GOV-001 · EU AI Act](https://santismm.com/en/governance/eu-ai-act): The EU AI Act is the European Union's comprehensive, risk-based law for artificial intelligence. It sorts AI systems into risk tiers — unacceptable (banned), high-risk (strict obligations), limited-risk (transparency duties) and minimal-risk — and adds specific obligations for general-purpose AI models. It applies extraterritorially to anyone placing AI on the EU market and phases in over several years, with penalties reaching up to 7% of global annual turnover for the most serious breaches. (evidence: industry_observation, confidence: high)
- [GOV-002 · ISO/IEC 42001](https://santismm.com/en/governance/iso-42001): ISO/IEC 42001:2023 is the first international, certifiable standard for an AI management system (AIMS). Like ISO 27001 for information security, it defines how an organization should establish, implement, maintain and continually improve the way it governs AI — through a policy, defined roles, risk and impact assessments, a set of controls, and a Plan-Do-Check-Act improvement cycle. It is voluntary and certifiable, giving organizations a recognized way to demonstrate responsible AI management. (evidence: industry_observation, confidence: high)
- [GOV-003 · NIST AI Risk Management Framework](https://santismm.com/en/governance/nist-ai-rmf): The NIST AI RMF 1.0 is a voluntary, widely-adopted framework for managing AI risk across the lifecycle. It is organized around four functions — Govern, Map, Measure and Manage — and a set of characteristics of trustworthy AI (valid and reliable, safe, secure and resilient, accountable and transparent, explainable, privacy-enhanced, and fair with harmful bias managed). A companion Generative AI Profile adapts it to GenAI risks. Unlike the EU AI Act it is not law, but it is a common backbone for operational AI governance. (evidence: industry_observation, confidence: high)
- [GOV-004 · Agentic AI Governance Checklist](https://santismm.com/en/governance/agentic-ai-governance-checklist): A practical, vendor-neutral checklist for governing agentic AI in the enterprise — translating the principles of the EU AI Act, ISO/IEC 42001 and NIST AI RMF into concrete controls you can implement in a harness. It covers human oversight, guardrails, audit logging, evaluation, access control, prompt-injection defence and incident response, and maps each control to the patterns and knowledge units that operationalize it. Use it as a readiness gate before letting an agent act in production. (evidence: industry_observation, confidence: medium)
- [GOV-005 · Enterprise AI Governance Framework](https://santismm.com/en/governance/enterprise-ai-governance-framework): An umbrella operating model for governing AI across an organization. It defines the principles, accountability (RACI), AI risk taxonomy, lifecycle gates and policy hierarchy that keep AI use lawful, safe and aligned with risk appetite. It harmonizes the EU AI Act, ISO/IEC 42001 and NIST AI RMF into one internal program — comply once, reuse everywhere — and composes the agentic governance checklist as its concrete control set. Use it to give every production AI system a named owner, a risk tier and a gate that can actually block a non-compliant deployment. (evidence: industry_observation, confidence: high)
- [GOV-006 · Audit Framework for Agentic Systems](https://santismm.com/en/governance/audit-framework-for-agentic-systems): A practical, vendor-neutral framework for making an agent auditable and for auditing it. It defines the evidence an independent reviewer needs — immutable, correlated traces of every decision and tool call, model and version provenance, evaluation reports, approval and incident records — and how to test controls and sample high-volume runs. Each evidence type maps to ISO/IEC 42001 and NIST AI RMF so an auditor can verify the agent stayed within its governed bounds. Use it to design auditability in from the start, not as an afterthought. (evidence: industry_observation, confidence: medium)
- [GOV-007 · Human Oversight and Accountability Policy](https://santismm.com/en/governance/human-oversight-and-accountability-policy): An operational policy that turns EU AI Act Article 14 human oversight into practice for agentic AI. It assigns a named accountable owner per agent, sets the oversight level (in-the-loop, on-the-loop, out-of-the-loop) by risk, and defines intervention, override and stop authority plus escalation paths. It requires overseers to be competent and have time to act, and it guards against rubber-stamping and automation bias. It exists to prevent two failures: the absent human and the token human who cannot actually understand, override, or answer for what the agent does. (evidence: industry_observation, confidence: high)

## Library

- [The Stopwatch and the Exam](https://articles.santismm.com/the-stopwatch-and-the-exam/): Static benchmarks did not die. They stopped being enough. The move from measuring capability to measuring agency, the harness problem underneath it, and where attention belongs now. Backed by a public catalog of 69 agentic benchmarks.
- [Cognitive Architecture and Emergent Phenomena in Advanced AI](https://santismm.substack.com/p/cognitive-architecture-and-emergent): A synthesis of current mechanistic and behavioral research on how cognitive structures and emergent behavior arise in modern AI systems.
- [The next generation of AI: Self-Improvement and Autonomous Learning](https://santismm.substack.com/p/the-next-generation-of-ai-self-improvement): Self-improvement and autonomous learning in AI — and the road to an intelligence explosion.
- [Vibe Coding in Enterprise Software](https://santismm.substack.com/p/vibe-coding-in-enterprise-software): AI-assisted software development in 2025: state of the art, risks and best practices for shipping it in real organizations.
- [The Rise of Agentic AI](https://santismm.substack.com/p/the-rise-of-agentic-ai): How agentic AI is revolutionizing business strategy — and the shift from do-it-yourself to do-it-for-me software.
- [Volando Alto con la Tecnología: una charla con Felipe Izquierdo, CTO de Sky Airlines](https://www.youtube.com/watch?v=el8E6AsMeUI): Charla con el CTO de Sky Airlines sobre transformación tecnológica, IA y experiencia de cliente en la aviación.
- [Transformando Negocios con IA Generativa](https://www.youtube.com/watch?v=Z3UBPqsx6XI): Conversación sobre cómo la IA Generativa transforma operaciones, productos y experiencia de cliente en la empresa.
- [Innovating Customer Experiences & Empowering Brands Through AI](https://www.youtube.com/watch?v=eC66YJDR8yA): DevReady Podcast (Ep. 108). A deep conversation on how conversational and generative AI are reshaping customer experience and the way brands operate.
- [The Era of Generative AI: How Large Multimodal Models are Reshaping Industries in 2024](https://medium.com/@santismm/the-era-of-generative-ai-how-large-multimodal-models-are-reshaping-industries-in-2024-fe62ae1678ab): Why 2024 became the year of the Large Multimodal Model (LMM), and what that means across industries.
- [How GenAI Impacts Business?](https://medium.com/@santismm/how-genai-impact-business-157238231948): A step-by-step guide to surf the innovation storm without sinking — how generative AI actually lands inside a business.
- [Rethinking the AI Pause: Potential Pitfalls and Unintended Consequences](https://www.linkedin.com/pulse/rethinking-ai-pause-potential-pitfalls-unintended-santiago): A measured take on the 2023 'AI pause' debate — what a moratorium would actually mean for innovation, safety and competition.
- [Unlock the Secret Language of Success: How GPT-4's Superpowers Are Changing the Game](https://www.linkedin.com/pulse/unlock-secret-language-success-how-gpt-4s-superpowers-santiago): An early look at GPT-4's qualitative leap and what its emergent capabilities meant for products and teams.

## Labs

- [Taxonomy of autonomous agents](https://santismm.com/en/labs/agent-taxonomy): An interactive taxonomy that encodes ~18 AI agents — from OpenAI, Anthropic, Google, Microsoft and Amazon to the open-source frontier — as vectors across 7 orthogonal axes, with a filterable table, an A×T×I governance-risk matrix and a machine-readable JSON API.
- [The professions most exposed to AI](https://santismm.com/en/labs/profesiones-ia): An interactive map of the 22 occupational groups, comparing AI's theoretical capability against its observed real-world use (as of June 2026), with a category ranking, the most-exposed occupations and the macro figures — rigorously sourced.
- [AI Agent Benchmarks](https://benchmarks.santismm.com/): A saturation tracker for AI-agent benchmarks: what they measure, SOTA, headroom, the human gap and research priority.
- [How demanding is each exam in the world?](https://santismm.com/labs/exigencia): An interactive benchmark comparing the academic demandingness of 18 university-entrance exams (PAES, PAU, IB, SAT, Gaokao, Suneung, JEE…) across 7 dimensions, with a world map, radars and ranking.
- [PAES vs PAU vs IB](https://santismm.com/labs/paes-pau-ib): An interactive benchmark comparing the academic demandingness of Chile's PAES, Spain's PAU and the IB Diploma across 6 cognitive dimensions and 8 subjects, with radar charts.
- [The money World Cup](https://santismm.com/en/labs/mundial-del-dinero): Which national team punches above its squad value at the 2026 World Cup. A live efficiency index — points vs €-value — team by team.
- [World Surf Atlas](https://santismm.com/en/sports/surf): An interactive atlas of 100+ world-class waves — wave direction (left / right / both), break type, level and month-by-month average wave height, plotted on a world map.
- [World Ski Atlas](https://santismm.com/en/sports/ski): 130+ ski resorts on a world map — skiable km, runs by difficulty, snow quality, monthly snowfall, temperature and sun, plus lift-pass prices.
- [World Bike Park Atlas](https://santismm.com/en/sports/ebike): 40+ lift-served MTB / e-bike parks worldwide — trails by difficulty, vertical drop, km, season window and day-pass prices, on a world map.

Every knowledge unit and pattern carries Evidence-First provenance — an evidence
level (production · simulation · benchmark · industry_observation · theoretical),
a confidence level and one or more source types — so you can weight each claim.
The full consolidated corpus, with that metadata, lives at /llms-full.txt.

## Feeds & discovery

- [Full corpus (llms-full.txt)](https://santismm.com/llms-full.txt)
- [RSS](https://santismm.com/feed.xml)
- [JSON Feed](https://santismm.com/feed.json)
- [Sitemap](https://santismm.com/sitemap.xml)
- [AI index (JSON)](https://santismm.com/ai-index.json)
- [For AI agents — every machine-readable surface, as followable links](https://santismm.com/en/for-agents)
- [MCP endpoint (native agent access, Streamable HTTP)](https://santismm.com/mcp) — POST JSON-RPC; see docs at https://santismm.com/en/mcp
- [MCP server manifest (registry)](https://santismm.com/.well-known/mcp/server.json)
- [Agent Skill — manifest (JSON)](https://santismm.com/skill.json) — how to install the skill and where its files are
- [Agent Skill — SKILL.md](https://santismm.com/skill/skill.md) — drop-in procedure for agents to query, cite and build with this corpus
- [Agent Skill — reference.md](https://santismm.com/skill/reference.md) — full MCP tool signatures and endpoint catalog
- [Knowledge graph (JSON)](https://santismm.com/api/graph.json)
- [Agent taxonomy (JSON)](https://santismm.com/api/agent-taxonomy.json)
- [Handbook API (JSON)](https://santismm.com/api/handbook)
- [Knowledge API (JSON)](https://santismm.com/api/knowledge)
- [Patterns API (JSON)](https://santismm.com/api/patterns)
- [Architectures API (JSON)](https://santismm.com/api/architectures)
- [Governance API (JSON)](https://santismm.com/api/governance)

## Profiles

- [LinkedIn](https://www.linkedin.com/in/santismm)
- [GitHub](https://github.com/santismm)
- [Medium](https://medium.com/@santismm)
- [YouTube](https://www.youtube.com/@santismorales)
- [X](https://x.com/santismorales)