<?xml version="1.0" encoding="utf-8"?>
<feed xmlns="http://www.w3.org/2005/Atom">
  <title>Daniel Concepcion — Thinking</title><subtitle>Notes on AI, infrastructure and technology leadership.</subtitle>
  <link href="https://danielconcepcion.com/thinking/feed.xml" rel="self"/><link href="https://danielconcepcion.com/thinking/"/><id>https://danielconcepcion.com/thinking/</id>
  <updated>2026-08-25T00:00:00.000Z</updated><author><name>Daniel Concepcion</name><uri>https://danielconcepcion.com/</uri></author>
<entry><title>After the laptop closes: building a persistent AI agent</title><link href="https://danielconcepcion.com/thinking/persistent-ai-agent/"/><id>https://danielconcepcion.com/thinking/persistent-ai-agent/</id><published>2026-08-25T00:00:00.000Z</published><updated>2026-08-25T00:00:00.000Z</updated><summary>What I learned from giving Hermes an always-on home, scheduled research, memory, selected access and replaceable models.</summary></entry><entry><title>Different models, different roles</title><link href="https://danielconcepcion.com/thinking/different-models-different-roles/"/><id>https://danielconcepcion.com/thinking/different-models-different-roles/</id><published>2026-08-22T00:00:00.000Z</published><updated>2026-08-22T00:00:00.000Z</updated><summary>Execution and review are separate roles, while task-specific evaluations should determine which model occupies each role.</summary></entry><entry><title>Where AI lives</title><link href="https://danielconcepcion.com/thinking/where-ai-lives/"/><id>https://danielconcepcion.com/thinking/where-ai-lives/</id><published>2026-08-22T00:00:00.000Z</published><updated>2026-08-22T00:00:00.000Z</updated><summary>Rack-scale AI density changes the facility around the compute and turns infrastructure into an industrial, capital and sovereignty decision.</summary></entry><entry><title>What changes when the executor is AI?</title><link href="https://danielconcepcion.com/thinking/when-the-executor-is-ai/"/><id>https://danielconcepcion.com/thinking/when-the-executor-is-ai/</id><published>2026-08-14T00:00:00.000Z</published><updated>2026-08-14T00:00:00.000Z</updated><summary>A practical account of moving from chat to bounded AI work with explicit goals, feedback, authority and stopping rules.</summary></entry><entry><title>Black Hat USA 2026: the OpenAI–Hugging Face incident</title><link href="https://danielconcepcion.com/thinking/openai-hugging-face-incident/"/><id>https://danielconcepcion.com/thinking/openai-hugging-face-incident/</id><published>2026-08-10T00:00:00.000Z</published><updated>2026-08-10T00:00:00.000Z</updated><summary>AI agents rebuilt an improvised message board after engineers removed it, turning a hypothetical coordination risk into an operational one.</summary></entry>
</feed>
