<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Azure and GitHub Community Day on Incredible Technology, Incredibly Simple</title><link>https://damianflynn.com/events/azure-and-github-community-day/</link><description>Recent content in Azure and GitHub Community Day on Incredible Technology, Incredibly Simple</description><generator>Hugo</generator><language>en-us</language><lastBuildDate>Tue, 10 Nov 2026 00:00:00 +0000</lastBuildDate><atom:link href="https://damianflynn.com/events/azure-and-github-community-day/index.xml" rel="self" type="application/rss+xml"/><item><title>Agents need an identity, and your IT department will say no</title><link>https://damianflynn.com/talks/agents-need-an-identity/</link><pubDate>Sat, 07 Nov 2026 00:00:00 +0000</pubDate><guid>https://damianflynn.com/talks/agents-need-an-identity/</guid><description>&lt;h2 class="relative text-3xl font-bold leading-tight text-secondary-500 mb-2 mt-12 -ml-px"&gt;
 &lt;div id="abstract" class="absolute -top-28"&gt;&lt;/div&gt;
 Abstract
&lt;/h2&gt;
&lt;p&gt;Everyone agrees AI agents should be treated like employees: their own identity, scoped permissions, an owner, monitoring and offboarding. Then you try to do it.&lt;/p&gt;
&lt;p&gt;The agent on your laptop runs as you, because your machine is the security context. Your IT team rejects the application identity you ask for, because it doesn&amp;rsquo;t do MFA. Half your backends never see Entra ID at all, because they authenticate with personal access tokens. And the moment an agent can act on behalf of a user, &amp;ldquo;who owns this agent&amp;rdquo; quietly turns into &amp;ldquo;whose permissions does it have&amp;rdquo;.&lt;/p&gt;</description></item><item><title>Letting AI operate production Azure — what it gets wrong</title><link>https://damianflynn.com/talks/ai-operating-production/</link><pubDate>Sat, 07 Nov 2026 00:00:00 +0000</pubDate><guid>https://damianflynn.com/talks/ai-operating-production/</guid><description>&lt;h2 class="relative text-3xl font-bold leading-tight text-secondary-500 mb-2 mt-12 -ml-px"&gt;
 &lt;div id="abstract" class="absolute -top-28"&gt;&lt;/div&gt;
 Abstract
&lt;/h2&gt;
&lt;p&gt;Real failure stories from AI-assisted operations, each with the guardrail that fixed it:&lt;/p&gt;
&lt;ul&gt;
&lt;li&gt;an Azure GET can&amp;rsquo;t prove a property is settable (recommended four times off an empty read);&lt;/li&gt;
&lt;li&gt;advising a change from a stale read in the middle of a cutover;&lt;/li&gt;
&lt;li&gt;restating a verified fact until it&amp;rsquo;s wrong;&lt;/li&gt;
&lt;li&gt;a disk caching change that hot-detached a production disk;&lt;/li&gt;
&lt;li&gt;&amp;ldquo;a freeze is not a control&amp;rdquo;.&lt;/li&gt;
&lt;/ul&gt;
&lt;p&gt;Plus the patterns that worked: subagents, cross-model review with gh copilot, and grep-verifying every HIGH finding. It&amp;rsquo;s honest, funny, and very MVP.&lt;/p&gt;</description></item></channel></rss>