<?xml version="1.0" encoding="UTF-8"?><rss version="2.0"><channel><title>Bloody Good Software · Insights</title><description>Case studies and field notes from building AI-assisted software for Australian small businesses.</description><link>https://bloodygoodsoftware.com/</link><language>en-au</language><item><title>The code took a day. The trust layer was the work.</title><link>https://bloodygoodsoftware.com/insights/sms-sender-id-register/</link><guid isPermaLink="true">https://bloodygoodsoftware.com/insights/sms-sender-id-register/</guid><description>Sending a text is the easy half. Getting your business name on it, in a way customers can trust, is the real work, and it is what Australia&apos;s SMS Sender ID Register protects.</description><pubDate>Mon, 15 Jun 2026 00:00:00 GMT</pubDate><category>SMS Sender ID Register</category><category>ACMA</category><category>branded SMS</category><category>Australian business</category></item><item><title>Why your AI gets worse the longer you talk to it</title><link>https://bloodygoodsoftware.com/insights/context-rot/</link><guid isPermaLink="true">https://bloodygoodsoftware.com/insights/context-rot/</guid><description>Long AI chats quietly degrade as they grow, even inside the stated limit. The cause is context rot, and a few plain habits fix most of it.</description><pubDate>Tue, 02 Jun 2026 00:00:00 GMT</pubDate><category>ai</category><category>context-rot</category><category>prompting</category></item><item><title>Putting AI to Work, Responsibly</title><link>https://bloodygoodsoftware.com/insights/responsibleai/</link><guid isPermaLink="true">https://bloodygoodsoftware.com/insights/responsibleai/</guid><description>How I dropped a large language model into a regulated Australian financial advice workflow without letting it generate the answers, decide the outcome, or leave the audit trail.</description><pubDate>Tue, 26 May 2026 00:00:00 GMT</pubDate><category>ai</category><category>compliance</category><category>case-study</category></item></channel></rss>