<?xml version="1.0" encoding="utf-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Sam Reghenzi Homepage</title><link>https://blog.r6i.it/</link><description/><atom:link href="https://blog.r6i.it/feeds/all.rss.xml" rel="self"/><lastBuildDate>Tue, 01 Sep 2026 09:56:00 +0200</lastBuildDate><item><title>The Ontology Was Always a Bounded Context</title><link>https://blog.r6i.it/ontology-was-always-a-bounded-context.html</link><description>&lt;p&gt;Two agents meet in a park.&lt;/p&gt;
&lt;p&gt;"I've just had a complaint from a customer," says the first.&lt;/p&gt;
&lt;p&gt;"What's a customer?" asks the second.&lt;/p&gt;
&lt;p&gt;"Anyone with an account."&lt;/p&gt;
&lt;p&gt;A third agent walks over.&lt;/p&gt;
&lt;p&gt;"That's not a customer," it says. "A customer is anyone with an active subscription."&lt;/p&gt;
&lt;p&gt;The three agents pause …&lt;/p&gt;</description><dc:creator xmlns:dc="http://purl.org/dc/elements/1.1/">samreghenzi</dc:creator><pubDate>Tue, 01 Sep 2026 09:56:00 +0200</pubDate><guid>tag:blog.r6i.it,2026-09-01:/ontology-was-always-a-bounded-context.html</guid><category>AI</category><category>AI</category><category>LLM</category><category>ontology</category><category>DDD</category><category>agents</category><category>MCP</category></item><item><title>The Guardrail Is Not in the Model</title><link>https://blog.r6i.it/ontologies-agentic-guardrails.html</link><description>&lt;p&gt;Frank Coyle gave &lt;a href="https://www.youtube.com/watch?v=Sir59K8ZDPU"&gt;a twenty-minute talk&lt;/a&gt; at the AI Engineer World's Fair that has been circulating for a few weeks now, and the argument in it is simple enough to state in one sentence: agentic systems need ontologies, because LLMs are probabilistic and probabilistic systems need logical guardrails.&lt;/p&gt;
&lt;p&gt;I think …&lt;/p&gt;</description><dc:creator xmlns:dc="http://purl.org/dc/elements/1.1/">samreghenzi</dc:creator><pubDate>Sun, 30 Aug 2026 14:16:00 +0200</pubDate><guid>tag:blog.r6i.it,2026-08-30:/ontologies-agentic-guardrails.html</guid><category>AI</category><category>AI</category><category>LLM</category><category>ontology</category><category>agents</category><category>python</category><category>claude</category></item><item><title>The Baseline That Refuses to Die</title><link>https://blog.r6i.it/classical-ml-2026-baselines.html</link><description>&lt;p&gt;For a decade, gradient-boosted trees owned tabular machine learning. In the last twelve months tabular foundation models finally — and decisively — beat them on TabArena. But the lesson practitioners are drawing ("stop using trees") is almost the opposite of what the evidence supports, because almost nobody runs the comparison correctly. A field guide to baselines, benchmarking discipline, and the GPU crunch in 2026.&lt;/p&gt;</description><dc:creator xmlns:dc="http://purl.org/dc/elements/1.1/">Sam Reghenzi</dc:creator><pubDate>Fri, 31 Jul 2026 09:30:00 +0200</pubDate><guid>tag:blog.r6i.it,2026-07-31:/classical-ml-2026-baselines.html</guid><category>AI</category><category>ml</category><category>tabular</category><category>foundation-models</category><category>xgboost</category><category>forecasting</category><category>benchmarks</category><category>gpu</category></item><item><title>Hardware Ate Software. But the Survivors Become Agents</title><link>https://blog.r6i.it/hardware-ate-software-survivors-become-agents.html</link><description>&lt;p&gt;A response to "Software Ate the World, Now Hardware Is Eating Software." The diagnosis is right — value is migrating and point software is being crushed. The prognosis is wrong. When the bottom of the stack commoditizes, value doesn't stay there. It climbs back up to the layer the thesis forgot: the agent.&lt;/p&gt;</description><dc:creator xmlns:dc="http://purl.org/dc/elements/1.1/">Sam Reghenzi</dc:creator><pubDate>Tue, 14 Jul 2026 09:30:00 +0200</pubDate><guid>tag:blog.r6i.it,2026-07-14:/hardware-ate-software-survivors-become-agents.html</guid><category>AI</category><category>llm</category><category>agents</category><category>hardware</category><category>inference</category><category>mcp</category><category>economics</category><category>stack</category></item><item><title>OAuth's Original Sin: Why MCP Inherited a Problem Designed for Humans</title><link>https://blog.r6i.it/oauth-original-sin-mcp-auth.html</link><description>&lt;p&gt;The third post in the MCP series. The first two argued &lt;em&gt;when&lt;/em&gt; MCP wins and loses. This one looks at the terrain MCP is supposed to win on — OAuth-bound integrations — and shows why even there, the ground is more fragile than it looks.&lt;/p&gt;</description><dc:creator xmlns:dc="http://purl.org/dc/elements/1.1/">Sam Reghenzi</dc:creator><pubDate>Fri, 19 Jun 2026 22:48:00 +0200</pubDate><guid>tag:blog.r6i.it,2026-06-19:/oauth-original-sin-mcp-auth.html</guid><category>AI</category><category>llm</category><category>mcp</category><category>oauth</category><category>security</category><category>claude</category><category>authentication</category></item><item><title>Skills vs MCP Servers: The Hidden Cost of Tools in LLM Workflows</title><link>https://blog.r6i.it/skills-vs-mcp-tokens-security.html</link><description>&lt;p&gt;Why CLI-based skills are often cheaper, safer, and more ergonomic than MCP servers — and the one case where MCP genuinely wins.&lt;/p&gt;</description><dc:creator xmlns:dc="http://purl.org/dc/elements/1.1/">Sam Reghenzi</dc:creator><pubDate>Sun, 07 Jun 2026 00:00:00 +0200</pubDate><guid>tag:blog.r6i.it,2026-06-07:/skills-vs-mcp-tokens-security.html</guid><category>AI</category><category>llm</category><category>mcp</category><category>skills</category><category>claude</category><category>tokens</category><category>security</category></item><item><title>MCP vs Skills: when to use which in professional LLM-assisted coding</title><link>https://blog.r6i.it/mcp-vs-skills-when-to-use-which-in-professional-llm-assisted-coding.html</link><description>&lt;p&gt;Anyone who has spent years doing systems integration immediately understands the appeal of MCP. The dream in this field has always been the same: a clean contract between systems, where every method is described, every parameter is typed, every response is predictable. No more 80-page PDFs read half-heartedly, no more …&lt;/p&gt;</description><dc:creator xmlns:dc="http://purl.org/dc/elements/1.1/">samreghenzi</dc:creator><pubDate>Sun, 10 May 2026 22:48:00 +0200</pubDate><guid>tag:blog.r6i.it,2026-05-10:/mcp-vs-skills-when-to-use-which-in-professional-llm-assisted-coding.html</guid><category>blog</category><category>AI</category><category>LLM</category><category>MCP</category><category>Python</category></item><item><title>Running Gemma 4 31B on an Apple Silicon Mac with Ollama</title><link>https://blog.r6i.it/running-gemma-4-31b-on-an-apple-silicon-mac-with-ollama.html</link><description>&lt;h2&gt;A practical configuration for a 32 GB M5 Mac that still needs to remain usable&lt;/h2&gt;
&lt;p&gt;Running large language models locally has become surprisingly practical on Apple Silicon. With a modern Mac, Ollama, and a carefully quantized GGUF model, it is possible to run models that only a short time ago …&lt;/p&gt;</description><dc:creator xmlns:dc="http://purl.org/dc/elements/1.1/">samreghenzi</dc:creator><pubDate>Sun, 26 Apr 2026 22:48:00 +0200</pubDate><guid>tag:blog.r6i.it,2026-04-26:/running-gemma-4-31b-on-an-apple-silicon-mac-with-ollama.html</guid><category>blog</category><category>AI</category><category>LLM</category><category>MCP</category><category>Python</category></item><item><title>Giving voice to Italian energy markets with artificial intelligence</title><link>https://blog.r6i.it/giving-voice-to-italian-energy-markets-with-artificial-intelligence.html</link><description>&lt;h2&gt;How an MCP server turns electricity exchange data into natural conversations&lt;/h2&gt;
&lt;p&gt;The Italian electricity market produces an enormous amount of data every day: hourly prices by zone, traded volumes, liquidity, the PUN (Prezzo Unico Nazionale — the national single price). Essential data for traders, energy analysts, industry journalists, and companies operating …&lt;/p&gt;</description><dc:creator xmlns:dc="http://purl.org/dc/elements/1.1/">samreghenzi</dc:creator><pubDate>Sun, 12 Apr 2026 22:48:00 +0200</pubDate><guid>tag:blog.r6i.it,2026-04-12:/giving-voice-to-italian-energy-markets-with-artificial-intelligence.html</guid><category>blog</category><category>AI</category><category>LLM</category><category>MCP</category><category>Python</category></item><item><title>Stop Putting Secrets in Your Repo. Seriously.</title><link>https://blog.r6i.it/stop-putting-secrets-in-your-repo-seriously.html</link><description>&lt;p&gt;OK. We've all been there. You're setting up a new microservice, you need a database connection string, a Keycloak client secret, an OCIR auth token — and before you know it, you're copy-pasting credentials into a &lt;code&gt;.tfvars&lt;/code&gt; file and telling yourself &lt;em&gt;"I'll fix this later"&lt;/em&gt;.&lt;/p&gt;
&lt;p&gt;Later never comes. Until it does …&lt;/p&gt;</description><dc:creator xmlns:dc="http://purl.org/dc/elements/1.1/">samreghenzi</dc:creator><pubDate>Fri, 27 Mar 2026 22:48:00 +0100</pubDate><guid>tag:blog.r6i.it,2026-03-27:/stop-putting-secrets-in-your-repo-seriously.html</guid><category>blog</category><category>devops</category><category>terraform</category><category>secops</category><category>dataengineering</category></item><item><title>Forecasting Time Series from a COVID-19 Red Zone</title><link>https://blog.r6i.it/forecasting-time-series-from-a-covid-19-red-zone.html</link><description>&lt;p&gt;Hello, dear citizen of the free world. I’m writing while I’m in a restricted lifestyle. On Monday night Italy’s prime minister decided that the entire country would be covered by restrictions that can be summarised as follows:&lt;/p&gt;
&lt;p&gt;“I stay at home”&lt;/p&gt;
&lt;p&gt;All travel was banned unless justified …&lt;/p&gt;</description><dc:creator xmlns:dc="http://purl.org/dc/elements/1.1/">samreghenzi</dc:creator><pubDate>Fri, 17 Apr 2020 22:48:00 +0200</pubDate><guid>tag:blog.r6i.it,2020-04-17:/forecasting-time-series-from-a-covid-19-red-zone.html</guid><category>blog</category><category>ML</category></item><item><title>I will ride like a gentlemen</title><link>https://blog.r6i.it/i-will-ride-like-a-gentlemen.html</link><description>&lt;h1&gt;I will ride like a gentlemen&lt;/h1&gt;
&lt;p&gt;On September 24th I will ride in the Distinguished Gentlemen Ride&lt;/p&gt;
&lt;p&gt;&lt;img alt="This is me an my bike at a recent Ducati Scrambler event" src="https://cdn-images-1.medium.com/max/2000/1*_2xCEsRJxtbLU8uSnI9n0w.jpeg"&gt;&lt;em&gt;This is me an my bike at a recent Ducati Scrambler event&lt;/em&gt;&lt;/p&gt;
&lt;p&gt;As a rider of a stylish Ducati Scrambler, I will join my local chapter of the DGR. It is a retro-themed …&lt;/p&gt;</description><dc:creator xmlns:dc="http://purl.org/dc/elements/1.1/">samreghenzi</dc:creator><pubDate>Mon, 17 Apr 2017 22:48:00 +0200</pubDate><guid>tag:blog.r6i.it,2017-04-17:/i-will-ride-like-a-gentlemen.html</guid><category>blog</category><category>motorcycle</category><category>lifestyle</category><category>personal</category></item></channel></rss>