<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:content="http://purl.org/rss/1.0/modules/content/">
  <channel>
    <title>Smara Blog</title>
    <link>https://smara.io/blog/</link>
    <description>Technical articles on AI agent memory, decay-scored retrieval, and building with Smara.</description>
    <language>en-us</language>
    <lastBuildDate>Tue, 06 May 2026 00:00:00 GMT</lastBuildDate>
    <atom:link href="https://smara.io/blog/feed.xml" rel="self" type="application/rss+xml"/>

    <item>
      <title>How to Give Gemini Persistent Memory with Smara</title>
      <link>https://smara.io/blog/gemini-memory-integration</link>
      <guid isPermaLink="true">https://smara.io/blog/gemini-memory-integration</guid>
      <pubDate>Tue, 06 May 2026 00:00:00 GMT</pubDate>
      <description>Give Gemini persistent memory across sessions using Smara's REST API. Three methods: direct REST calls, Gemini function calling, and Google ADK integration.</description>
      <author>sri@smara.io (Sri Panchavati)</author>
    </item>

    <item>
      <title>Smara vs Building Your Own Memory with pgvector</title>
      <link>https://smara.io/blog/smara-vs-diy-pgvector</link>
      <guid isPermaLink="true">https://smara.io/blog/smara-vs-diy-pgvector</guid>
      <pubDate>Tue, 06 May 2026 00:00:00 GMT</pubDate>
      <description>Should you build your own AI memory system with pgvector or use Smara? An honest comparison with LoCoMo benchmark scores, cost analysis, and architecture trade-offs.</description>
      <author>sri@smara.io (Sri Panchavati)</author>
    </item>

    <item>
      <title>How We Scored 92.2% on LoCoMo — The Hardest AI Memory Benchmark</title>
      <link>https://smara.io/blog/locomo-benchmark</link>
      <guid isPermaLink="true">https://smara.io/blog/locomo-benchmark</guid>
      <pubDate>Tue, 06 May 2026 00:00:00 GMT</pubDate>
      <description>LoCoMo tests whether AI memory systems can recall facts from long, multi-session conversations. Here's how we went from 62% to 92.2% across 9 versions.</description>
      <author>sri@smara.io (Sri Panchavati)</author>
    </item>

    <item>
      <title>Smara SDK Quickstart: Add AI Memory in 5 Minutes</title>
      <link>https://smara.io/blog/smara-sdk-quickstart</link>
      <guid isPermaLink="true">https://smara.io/blog/smara-sdk-quickstart</guid>
      <pubDate>Sat, 26 Apr 2026 00:00:00 GMT</pubDate>
      <description>Complete quickstart guide with full working examples in JavaScript, Python, and cURL. Store, search, list, and delete memories.</description>
      <author>sri@smara.io (Sri Panchavati)</author>
    </item>

    <item>
      <title>Smara vs Mem0: Which AI Memory API Should You Use in 2026?</title>
      <link>https://smara.io/blog/smara-vs-mem0</link>
      <guid isPermaLink="true">https://smara.io/blog/smara-vs-mem0</guid>
      <pubDate>Sat, 26 Apr 2026 00:00:00 GMT</pubDate>
      <description>A detailed comparison of pricing, features, architecture, and developer experience between Smara and Mem0.</description>
      <author>sri@smara.io (Sri Panchavati)</author>
    </item>

    <item>
      <title>How Ebbinghaus Decay Curves Make AI Memory Actually Useful</title>
      <link>https://smara.io/blog/ebbinghaus-ai-memory</link>
      <guid isPermaLink="true">https://smara.io/blog/ebbinghaus-ai-memory</guid>
      <pubDate>Sun, 27 Apr 2026 00:00:00 GMT</pubDate>
      <description>Why flat memory retrieval fails as memories accumulate, how the 1885 forgetting curve solves it, and how Smara implements decay-scored retrieval.</description>
      <author>sri@smara.io (Sri Panchavati)</author>
    </item>

    <item>
      <title>Building AI Agents with Persistent Memory: A Complete Guide</title>
      <link>https://smara.io/blog/ai-agents-persistent-memory</link>
      <guid isPermaLink="true">https://smara.io/blog/ai-agents-persistent-memory</guid>
      <pubDate>Sun, 27 Apr 2026 00:00:00 GMT</pubDate>
      <description>Step-by-step tutorial: build an AI agent that remembers across conversations. Covers storing, searching, agent-scoped memory, graph connections, and team memory.</description>
      <author>sri@smara.io (Sri Panchavati)</author>
    </item>

    <item>
      <title>The Best Mem0 Alternatives for AI Memory in 2026</title>
      <link>https://smara.io/blog/best-mem0-alternatives</link>
      <guid isPermaLink="true">https://smara.io/blog/best-mem0-alternatives</guid>
      <pubDate>Sun, 27 Apr 2026 00:00:00 GMT</pubDate>
      <description>Comparing Smara, Zep, Letta (MemGPT), agentmemory, and custom Postgres+pgvector for AI memory.</description>
      <author>sri@smara.io (Sri Panchavati)</author>
    </item>

    <item>
      <title>Add Memory to Claude Code in 30 Seconds with MCP</title>
      <link>https://smara.io/blog/claude-code-memory-mcp</link>
      <guid isPermaLink="true">https://smara.io/blog/claude-code-memory-mcp</guid>
      <pubDate>Sat, 26 Apr 2026 00:00:00 GMT</pubDate>
      <description>Install the Smara MCP server and give Claude Code persistent memory across sessions. Three steps, no code changes.</description>
      <author>sri@smara.io (Sri Panchavati)</author>
    </item>

  </channel>
</rss>
