<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom">
  <channel>
    <title>Murat Durak — Signal &amp; Essays</title>
    <link>https://muratdurak.com</link>
    <description>Field notes, teardowns and frameworks from live AI deployments.</description>
    <language>en</language>
    <atom:link href="https://muratdurak.com/rss.xml" rel="self" type="application/rss+xml" />
    <lastBuildDate>Wed, 23 Sep 2026 14:50:00 GMT</lastBuildDate>
    <item>
      <title>Signal No. 6 — The cost line nobody forecasts</title>
      <link>https://muratdurak.com/signal/archive#issue-6</link>
      <guid isPermaLink="true">https://muratdurak.com/signal/archive#issue-6</guid>
      <description>Inference is rarely the expensive part. Review, correction and escalation are — and they land on a team that never signed up for them.</description>
      <category>Signal</category>
      <pubDate>Sun, 20 Sep 2026 09:00:00 GMT</pubDate>
    </item>
    <item>
      <title>Signal No. 5 — Why the pilot passed and production failed</title>
      <link>https://muratdurak.com/signal/archive#issue-5</link>
      <guid isPermaLink="true">https://muratdurak.com/signal/archive#issue-5</guid>
      <description>Pilots are run by the people who want them to work. Production is run by everybody else.</description>
      <category>Signal</category>
      <pubDate>Sun, 13 Sep 2026 09:00:00 GMT</pubDate>
    </item>
    <item>
      <title>Signal No. 4 — One named owner, or nothing</title>
      <link>https://muratdurak.com/signal/archive#issue-4</link>
      <guid isPermaLink="true">https://muratdurak.com/signal/archive#issue-4</guid>
      <description>The fastest predictor of a deployment surviving its first bad week is whose phone rings at 2am.</description>
      <category>Signal</category>
      <pubDate>Sun, 06 Sep 2026 09:00:00 GMT</pubDate>
    </item>
    <item>
      <title>Signal No. 3 — Six weeks to two</title>
      <link>https://muratdurak.com/signal/archive#issue-3</link>
      <guid isPermaLink="true">https://muratdurak.com/signal/archive#issue-3</guid>
      <description>A teardown of a cycle-time reduction that had almost nothing to do with the model choice.</description>
      <category>Signal</category>
      <pubDate>Sun, 30 Aug 2026 09:00:00 GMT</pubDate>
    </item>
    <item>
      <title>Signal No. 2 — What a growth system actually does</title>
      <link>https://muratdurak.com/signal/archive#issue-2</link>
      <guid isPermaLink="true">https://muratdurak.com/signal/archive#issue-2</guid>
      <description>Not campaigns. A growth system is the set of decisions you no longer have to make by hand each week.</description>
      <category>Signal</category>
      <pubDate>Sun, 23 Aug 2026 09:00:00 GMT</pubDate>
    </item>
    <item>
      <title>The Deployment Ladder: Five Rungs from Pilot to Production</title>
      <link>https://muratdurak.com/thinking/the-deployment-ladder</link>
      <guid isPermaLink="true">https://muratdurak.com/thinking/the-deployment-ladder</guid>
      <description>A working model for moving enterprise AI out of the demo stage: evidence, ownership, evaluation, integration and standardisation, taken one rung at a time.</description>
      <category>Strategy</category>
      <pubDate>Thu, 20 Aug 2026 09:00:00 GMT</pubDate>
    </item>
    <item>
      <title>Signal No. 1 — What I see inside AI deployments</title>
      <link>https://muratdurak.com/signal/archive#issue-1</link>
      <guid isPermaLink="true">https://muratdurak.com/signal/archive#issue-1</guid>
      <description>Why this briefing exists, and the one question I ask in the first meeting of every engagement.</description>
      <category>Signal</category>
      <pubDate>Sun, 16 Aug 2026 09:00:00 GMT</pubDate>
    </item>
    <item>
      <title>Teardown: Six Weeks to Two</title>
      <link>https://muratdurak.com/thinking/teardown-six-weeks-to-two</link>
      <guid isPermaLink="true">https://muratdurak.com/thinking/teardown-six-weeks-to-two</guid>
      <description>An anonymised case teardown — problem, intervention, result — from a mandate that cut a customer operations workflow from six weeks to two.</description>
      <category>Case Teardown</category>
      <pubDate>Wed, 12 Aug 2026 09:00:00 GMT</pubDate>
    </item>
    <item>
      <title>What Growth Systems Actually Do</title>
      <link>https://muratdurak.com/thinking/what-growth-systems-actually-do</link>
      <guid isPermaLink="true">https://muratdurak.com/thinking/what-growth-systems-actually-do</guid>
      <description>Growth is not a campaign function. It is an architecture question: where revenue leaks, who owns each handoff, and which decisions can be made without a meeting.</description>
      <category>Growth</category>
      <pubDate>Tue, 14 Jul 2026 09:00:00 GMT</pubDate>
    </item>
  </channel>
</rss>