<?xml version="1.0" encoding="UTF-8"?><rss version="2.0"><channel><title>Epithre Blog</title><description>Notes from Epithre on the AI we build and run ourselves in Indonesia.</description><link>https://www.epithre.com/</link><language>en</language><item><title>Introducing Strata</title><link>https://www.epithre.com/blog/introducing-strata/</link><guid isPermaLink="true">https://www.epithre.com/blog/introducing-strata/</guid><description>Strata is Epithre&apos;s language model for Indonesian government work: trained on the country&apos;s regulations and run without sending data outside.</description><pubDate>Fri, 29 May 2026 10:00:00 GMT</pubDate><category>newsroom</category><category>strata</category><category>sovereign ai</category><category>government</category></item><item><title>Sovereign AI is more than where the server sits</title><link>https://www.epithre.com/blog/sovereign-ai-is-more-than-server-location/</link><guid isPermaLink="true">https://www.epithre.com/blog/sovereign-ai-is-more-than-server-location/</guid><description>Many equate AI sovereignty with putting servers in-country. The reality is more complicated, and more decisive, than that.</description><pubDate>Fri, 29 May 2026 09:00:00 GMT</pubDate><category>kedaulatan</category><category>sovereign ai</category><category>data sovereignty</category><category>government</category></item><item><title>How Strata reasons about regulatory hierarchy</title><link>https://www.epithre.com/blog/how-strata-reasons-about-regulatory-hierarchy/</link><guid isPermaLink="true">https://www.epithre.com/blog/how-strata-reasons-about-regulatory-hierarchy/</guid><description>A question that looks trivial, which rule wins when two conflict, is exactly what shows why regulatory reasoning takes more than a big model.</description><pubDate>Fri, 29 May 2026 08:30:00 GMT</pubDate><category>showcase</category><category>strata</category><category>capability</category><category>regulation</category></item><item><title>Why we built our own AI benchmark</title><link>https://www.epithre.com/blog/why-we-built-our-own-ai-benchmark/</link><guid isPermaLink="true">https://www.epithre.com/blog/why-we-built-our-own-ai-benchmark/</guid><description>A model can top the popular benchmarks and still be useless to a ministry. Here is why we built our own evaluation for Indonesian government work.</description><pubDate>Thu, 28 May 2026 12:00:00 GMT</pubDate><category>riset</category><category>research</category><category>evaluation</category><category>benchmark</category></item><item><title>About this blog</title><link>https://www.epithre.com/blog/about-this-blog/</link><guid isPermaLink="true">https://www.epithre.com/blog/about-this-blog/</guid><description>Notes from Epithre on the AI we build and run ourselves in Indonesia: models, research, sovereignty, and how it is used.</description><pubDate>Thu, 28 May 2026 00:00:00 GMT</pubDate><category>newsroom</category><category>epithre</category><category>sovereign ai</category></item></channel></rss>