<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:content="http://purl.org/rss/1.0/modules/content/"><channel><title>Calyntro Blog</title><link>https://calyntro.com/blog/</link><description>Recent content on Calyntro Blog</description><generator>Hugo</generator><language>en-us</language><lastBuildDate>Tue, 06 Oct 2026 09:00:00 +0200</lastBuildDate><atom:link href="https://calyntro.com/blog/index.xml" rel="self" type="application/rss+xml"/><item><title>SDD Beyond the Prototype, Part 1: Where Spec-Driven Development Breaks Once More Than One Person Is Writing</title><link>https://calyntro.com/blog/architecture/spec-driven-development/part-1-where-sdd-breaks/</link><pubDate>Fri, 28 Aug 2026 09:00:00 +0200</pubDate><guid>https://calyntro.com/blog/architecture/spec-driven-development/part-1-where-sdd-breaks/</guid><description>A working spec-driven development prototype tells you almost nothing about what happens once more than one person and one team depend on the same spec.</description></item><item><title>The Field Took Twenty Minutes</title><link>https://calyntro.com/blog/architecture/refactoring/the-field-took-twenty-minutes/</link><pubDate>Tue, 30 Jun 2026 00:00:00 +0200</pubDate><guid>https://calyntro.com/blog/architecture/refactoring/the-field-took-twenty-minutes/</guid><description>Why a trivial feature takes three days — and what that says about your architecture, not your skills.</description></item><item><title>Your Codebase Is Trying to Tell You Something</title><link>https://calyntro.com/blog/architecture/refactoring/your-codebase-is-trying-to-tell-you-something/</link><pubDate>Wed, 08 Jul 2026 00:00:00 +0200</pubDate><guid>https://calyntro.com/blog/architecture/refactoring/your-codebase-is-trying-to-tell-you-something/</guid><description>Change coupling is the hidden dependency your import graph will never show — but your git history records every time it happens.</description></item><item><title>The Hotspot Trap: Complexity × Churn as a Structural Signal</title><link>https://calyntro.com/blog/architecture/refactoring/the-hotspot-trap/</link><pubDate>Wed, 22 Jul 2026 00:00:00 +0000</pubDate><guid>https://calyntro.com/blog/architecture/refactoring/the-hotspot-trap/</guid><description>Complexity alone is fine. Churn alone is fine. The dangerous combination is both — in the same file. Here&amp;#39;s what that signal actually means.</description></item><item><title>Calyntro 1.16 — AI Briefing ready for shared instances</title><link>https://calyntro.com/blog/posts/2026-10-06-calyntro-1-16-ai-briefing/</link><pubDate>Tue, 06 Oct 2026 09:00:00 +0200</pubDate><guid>https://calyntro.com/blog/posts/2026-10-06-calyntro-1-16-ai-briefing/</guid><description>Navigation restructured, AI Briefing redesigned and renamed, monthly request cap for shared instances, and OpenAI as an alternative provider.</description></item><item><title>Tests Don't Care Who Typed the Code</title><link>https://calyntro.com/blog/posts/2026-07-03-tests-dont-care-who-typed-the-code/</link><pubDate>Wed, 01 Jul 2026 09:00:00 +0200</pubDate><guid>https://calyntro.com/blog/posts/2026-07-03-tests-dont-care-who-typed-the-code/</guid><description>AI-generated code has the same failure modes as human code. But there&amp;#39;s a subtler trap: when one model writes the spec, the test, and the implementation, it can validate its own mistakes. Here&amp;#39;s how to break that loop.</description></item><item><title>Nobody Chose BSUF. You Just Followed the Path of Least Resistance.</title><link>https://calyntro.com/blog/posts/2026-06-30-bsuf-temptation/</link><pubDate>Tue, 30 Jun 2026 00:00:00 +0000</pubDate><guid>https://calyntro.com/blog/posts/2026-06-30-bsuf-temptation/</guid><description>AI doesn&amp;#39;t bring back Big Spec Up Front as a philosophy. It brings it back as a default — one reasonable prompt at a time.</description></item><item><title>Two Doors, One Gate: Navigating Governance Beyond EDD</title><link>https://calyntro.com/blog/posts/two_doors_one_gate/</link><pubDate>Sun, 28 Jun 2026 00:00:00 +0000</pubDate><guid>https://calyntro.com/blog/posts/two_doors_one_gate/</guid><description>Onboarding guardrails and power-user friction look like the same problem. They aren&amp;#39;t. Here is why mixing awareness and governance at the developer level repeats a mistake we already made at the merge gate.</description></item><item><title>What Your Git History Reveals About Team Alignment</title><link>https://calyntro.com/blog/posts/2026-06-17-blog_team_alignment/</link><pubDate>Tue, 23 Jun 2026 00:00:00 +0000</pubDate><guid>https://calyntro.com/blog/posts/2026-06-17-blog_team_alignment/</guid><description>Your org chart shows who should own what. Your git history shows who actually does. They&amp;#39;re rarely the same — and the gap is readable. Here&amp;#39;s how.</description></item><item><title>EDD Closes the Loop — But Only Half of It</title><link>https://calyntro.com/blog/posts/2026-06-20-edd-closes-half-the-loop/</link><pubDate>Sat, 20 Jun 2026 00:00:00 +0000</pubDate><guid>https://calyntro.com/blog/posts/2026-06-20-edd-closes-half-the-loop/</guid><description>Expectation-Driven Development is a genuine step forward for AI-assisted development. But it solves the wrong half of the problem. Here is what the other half requires.</description></item><item><title>Knowledge Risk: From Metric to Recommended Action</title><link>https://calyntro.com/blog/posts/2026-06-10-knowledge-risk-recommended-actions/</link><pubDate>Tue, 16 Jun 2026 00:00:00 +0000</pubDate><guid>https://calyntro.com/blog/posts/2026-06-10-knowledge-risk-recommended-actions/</guid><description>Most tools tell you that knowledge is concentrated. Calyntro tells you what to do about it — and why some risks are more urgent than they appear when you combine ownership data with change coupling.</description></item><item><title>Bad Sprints Start Before the Sprint</title><link>https://calyntro.com/blog/posts/2026-06-14-bad-sprints-start-before-the-sprint/</link><pubDate>Sun, 14 Jun 2026 00:00:00 +0000</pubDate><guid>https://calyntro.com/blog/posts/2026-06-14-bad-sprints-start-before-the-sprint/</guid><description>Teams blame the board when sprints go sideways. The actual problem is almost always in refinement — and it starts long before the sprint does.</description></item><item><title>Conway's Law Is Already in Your Commit History</title><link>https://calyntro.com/blog/posts/2026-06-10-blog_conway_s_law/</link><pubDate>Wed, 10 Jun 2026 00:00:00 +0000</pubDate><guid>https://calyntro.com/blog/posts/2026-06-10-blog_conway_s_law/</guid><description>Conway&amp;#39;s Law has been treated as a design principle for 58 years. It&amp;#39;s time to treat it as something measurable — and your Git history already has the data.</description></item><item><title>Was Ihre Git-Historie über Ihr Team verrät — und warum kaum jemand hinschaut</title><link>https://calyntro.com/blog/posts/2026-06-07-5-warnsignale-git-historie/</link><pubDate>Sun, 07 Jun 2026 00:00:00 +0000</pubDate><guid>https://calyntro.com/blog/posts/2026-06-07-5-warnsignale-git-historie/</guid><description>5 Risikosignale in jeder Codebase — Hotspots, Wissenssilos, Change Coupling. Sie stehen bereits in Ihrer Git-Historie. Kaum ein Team liest sie systematisch.</description></item><item><title>Change Coupling: How Git History Reveals Conway's Law Violations</title><link>https://calyntro.com/blog/posts/2026-06-03-change-coupling-conways-law/</link><pubDate>Wed, 03 Jun 2026 00:00:00 +0000</pubDate><guid>https://calyntro.com/blog/posts/2026-06-03-change-coupling-conways-law/</guid><description>We ran change coupling analysis on MongoDB&amp;#39;s open-source repository. 208 module pairs cross team boundaries in the commit history — and the direction of that coupling tells you something no architecture diagram can.</description></item><item><title>How to Measure Bus Factor in Your Software Team</title><link>https://calyntro.com/blog/posts/2026-06-02-how-to-measure-bus-factor/</link><pubDate>Tue, 02 Jun 2026 00:00:00 +0000</pubDate><guid>https://calyntro.com/blog/posts/2026-06-02-how-to-measure-bus-factor/</guid><description>Bus factor tells you how many people you can lose before a project stalls. Most teams guess. Here&amp;#39;s how to measure it directly from Git history — no surveys, no interviews.</description></item><item><title>The Last Mile Problem in AI-Assisted Development</title><link>https://calyntro.com/blog/posts/2026-05-31-last_mile_problem/</link><pubDate>Sun, 31 May 2026 00:00:00 +0000</pubDate><guid>https://calyntro.com/blog/posts/2026-05-31-last_mile_problem/</guid><description>AI reviews can flag issues. But flags are not gates. The last mile — from finished code to trusted repository — is still unsolved. Here is why, and what actually helps.</description></item><item><title>SDD and the Missing Half of the Answer</title><link>https://calyntro.com/blog/posts/2026-05-28-ssd_arguments/</link><pubDate>Thu, 28 May 2026 00:00:00 +0000</pubDate><guid>https://calyntro.com/blog/posts/2026-05-28-ssd_arguments/</guid><description>Adam Tornhill is right that Spec-Driven Development replays the MDA failure. But there&amp;#39;s a second failure mode nobody is talking about: knowledge evaporation.</description></item><item><title>What Calyntro Measures — And Why Standard Tools Miss It</title><link>https://calyntro.com/blog/posts/2026-05-26-calyntro_week4/</link><pubDate>Tue, 26 May 2026 00:00:00 +0000</pubDate><guid>https://calyntro.com/blog/posts/2026-05-26-calyntro_week4/</guid><description>Most ownership tools answer who last touched a file. Calyntro asks a harder question: who would be left holding it if the person who built it walked out the door tomorrow?</description></item><item><title>From Chaos to Coherence: What AI Cannot Do for Your Architecture</title><link>https://calyntro.com/blog/posts/2026-05-25-architecture_too_late/</link><pubDate>Mon, 25 May 2026 00:00:00 +0000</pubDate><guid>https://calyntro.com/blog/posts/2026-05-25-architecture_too_late/</guid><description>AI builds fast. But coherence is the architect&amp;#39;s job — not the AI&amp;#39;s. A concrete story of PoC chaos, a full reset, and what only humans can define in a system.</description></item><item><title>The Review Is Too Late</title><link>https://calyntro.com/blog/posts/2026-05-20-the_review_is_too_late/</link><pubDate>Wed, 20 May 2026 00:00:00 +0000</pubDate><guid>https://calyntro.com/blog/posts/2026-05-20-the_review_is_too_late/</guid><description>Everyone is optimising the code review. But the review isn&amp;#39;t where quality gets decided — it&amp;#39;s where quality gets caught. In the age of AI-generated code, that distinction matters more than ever.</description></item><item><title>Why We Code in Threes</title><link>https://calyntro.com/blog/posts/programming_in_threes/</link><pubDate>Fri, 15 May 2026 00:00:00 +0000</pubDate><guid>https://calyntro.com/blog/posts/programming_in_threes/</guid><description>Two developers, one AI agent — and the reasons it works better than going alone.</description></item><item><title>What We Found When We Analysed MongoDB's Codebase</title><link>https://calyntro.com/blog/posts/2026-05-13-if-it-happens/</link><pubDate>Wed, 13 May 2026 08:00:00 +0200</pubDate><guid>https://calyntro.com/blog/posts/2026-05-13-if-it-happens/</guid><description>A file can look shared today but have been built entirely by someone who left last year. We ran that analysis on MongoDB — and the results are worth paying attention to.</description></item><item><title>The Invisible Risk in Your Codebase</title><link>https://calyntro.com/blog/posts/2026-04-30-invisible_risks_in_codebase/</link><pubDate>Thu, 30 Apr 2026 08:00:00 +0200</pubDate><guid>https://calyntro.com/blog/posts/2026-04-30-invisible_risks_in_codebase/</guid><description>Most teams know that critical knowledge is concentrated in a few people. Almost none know exactly where. That&amp;#39;s not a knowledge transfer problem — it&amp;#39;s a visibility problem.</description></item><item><title>Why AI Cannot Handle Full Specifications – and How Architecture with AI Really Works</title><link>https://calyntro.com/blog/posts/why-ai-cannot-handle-full-specifications/</link><pubDate>Mon, 20 Apr 2026 00:00:00 +0000</pubDate><guid>https://calyntro.com/blog/posts/why-ai-cannot-handle-full-specifications/</guid><description>Why AI fails with full-system specifications, how to structure systems in an AI-friendly way, and why component-based architecture is the key.</description></item><item><title>The Parallelization Paradox: Why Speed is Killing Your Agility</title><link>https://calyntro.com/blog/posts/2026-04-10-the-parallelization-paradox/</link><pubDate>Fri, 10 Apr 2026 10:00:00 +0200</pubDate><guid>https://calyntro.com/blog/posts/2026-04-10-the-parallelization-paradox/</guid><description>Why optimizing for simultaneous output with coding agents often leads back to a rigid Waterfall mindset.</description></item><item><title>AI + TDD: A Shortcut to the Goal or a Loss of Insight?</title><link>https://calyntro.com/blog/posts/ai-tdd-shortcut-or-loss-of-insight/</link><pubDate>Tue, 21 Apr 2026 09:00:00 +0200</pubDate><guid>https://calyntro.com/blog/posts/ai-tdd-shortcut-or-loss-of-insight/</guid><description>&lt;p&gt;Test-Driven Development has always been slightly misunderstood — even by people who practice it.&lt;/p&gt;
&lt;p&gt;The name doesn&amp;rsquo;t help. &amp;ldquo;Test-Driven&amp;rdquo; sounds like it&amp;rsquo;s primarily about tests. Coverage metrics. Regression safety. The QA team&amp;rsquo;s peace of mind. But anyone who has worked seriously with TDD, or spent time with practitioners like Emily Bache, knows that tests are almost a side effect. The real output is &lt;em&gt;understanding&lt;/em&gt;.&lt;/p&gt;
&lt;p&gt;TDD, done well, is a method for thinking your way through a problem one small step at a time. You don&amp;rsquo;t start with a complete picture of the solution. You start with the smallest possible question: &lt;em&gt;what is the simplest behavior this code should exhibit?&lt;/em&gt; You write a test for that. You make it pass. And in the process of making it pass, you learn something — about the problem, about your assumptions, about the design that is quietly trying to emerge.&lt;/p&gt;</description></item><item><title>Software Engineering: The Art of Thinking Out Loud (with AI)</title><link>https://calyntro.com/blog/posts/software-engineering-thinking-out-loud-with-ai/</link><pubDate>Tue, 14 Apr 2026 09:00:00 +0200</pubDate><guid>https://calyntro.com/blog/posts/software-engineering-thinking-out-loud-with-ai/</guid><description>&lt;p&gt;A colleague said something to me recently that I keep coming back to:&lt;/p&gt;
&lt;p&gt;&lt;em&gt;&amp;ldquo;Often, by the time you&amp;rsquo;ve finished articulating a complex problem for the AI, you&amp;rsquo;ve already solved it yourself.&amp;rdquo;&lt;/em&gt;&lt;/p&gt;
&lt;p&gt;It sounds almost like a joke. You open a chat window, start typing out your problem in careful detail — and somewhere in the middle of the second paragraph, the answer appears. Not from the AI. From you.&lt;/p&gt;
&lt;p&gt;If you&amp;rsquo;ve worked with LLMs seriously, you&amp;rsquo;ve probably experienced this. And I think it points to something important about what is actually changing in our craft — something that goes beyond the usual conversation about automation and job displacement.&lt;/p&gt;</description></item><item><title>AI in Software Engineering: Between Vision and Craftsmanship</title><link>https://calyntro.com/blog/posts/ai-in-software-engineering-vision-and-craftsmanship/</link><pubDate>Mon, 07 Apr 2025 00:00:00 +0000</pubDate><guid>https://calyntro.com/blog/posts/ai-in-software-engineering-vision-and-craftsmanship/</guid><description>&lt;p&gt;A question recently surfaced in my LinkedIn feed that I haven&amp;rsquo;t been able to shake:
&lt;em&gt;How will the next generation of software developers gain experience when AI takes over the writing?&lt;/em&gt;&lt;/p&gt;
&lt;p&gt;The analogy offered was that of a conductor — someone who doesn&amp;rsquo;t need to play the violin to judge whether the orchestra is in harmony. It&amp;rsquo;s a compelling metaphor. But I think it lets us off the hook too easily.&lt;/p&gt;</description></item></channel></rss>