<?xml version="1.0" encoding="UTF-8"?><rss version="2.0"><channel><title>Liath Blog</title><description>Insights on AI agent memory, programmable retrieval, and building Liath — the SQLite for AI agents.</description><link>https://liath.incredlabs.com/</link><item><title>Why Agents Need Programmable Memory</title><link>https://liath.incredlabs.com/blog/why-agents-need-programmable-memory/</link><guid isPermaLink="true">https://liath.incredlabs.com/blog/why-agents-need-programmable-memory/</guid><description>Fixed vector-database APIs give agents one retrieval strategy. Programmable memory lets an agent write the retrieval logic itself — safely — and that changes what agents can remember.</description><pubDate>Sat, 20 Jun 2026 00:00:00 GMT</pubDate><author>Dipankar Sarkar</author></item><item><title>Liath vs Vector Databases</title><link>https://liath.incredlabs.com/blog/liath-vs-vector-databases/</link><guid isPermaLink="true">https://liath.incredlabs.com/blog/liath-vs-vector-databases/</guid><description>Vector databases store and search embeddings behind a fixed API. Liath is embedded, programmable memory: agents query vectors, key-value data, and embeddings together with sandboxed Lua.</description><pubDate>Thu, 18 Jun 2026 00:00:00 GMT</pubDate><author>Dipankar Sarkar</author></item></channel></rss>