<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom">
  <channel>
    <title>Orqen Blog</title>
    <link>https://orqen.app/blog/</link>
    <description>Technical articles on LLM agent optimization, token cost reduction, and production AI.</description>
    <language>en</language>
    <atom:link href="https://orqen.app/blog/rss.xml/" rel="self" type="application/rss+xml"/>
    <item>
      <title><![CDATA[Tokenmaxxing Is Dead: How to Cut Your LLM Bill 50–70%]]></title>
      <link>https://orqen.app/blog/tokenmaxxing-is-dead/</link>
      <guid isPermaLink="true">https://orqen.app/blog/tokenmaxxing-is-dead/</guid>
      <pubDate>Sat, 30 May 2026 00:00:00 GMT</pubDate>
      <description><![CDATA[Tokenmaxxing backfired — big bills, no ROI. LLM cost optimization cuts the tokens your agent never needed: reduce LLM costs 50–70% without changing your code.]]></description>
    </item>
    <item>
      <title><![CDATA[Agent Called a Tool You Didn't Send? Fix Recall Misses]]></title>
      <link>https://orqen.app/blog/agent-tool-recall-misses-recovery/</link>
      <guid isPermaLink="true">https://orqen.app/blog/agent-tool-recall-misses-recovery/</guid>
      <pubDate>Thu, 21 May 2026 00:00:00 GMT</pubDate>
      <description><![CDATA[Per-turn tool routing can drop a tool the model still needs. How recall@K catches misses, how session recovery responds, and what Orqen can — and cannot — fix.]]></description>
    </item>
    <item>
      <title><![CDATA[Turn 47 Hit the Context Window. Now What?]]></title>
      <link>https://orqen.app/blog/turn-47-context-window-limits/</link>
      <guid isPermaLink="true">https://orqen.app/blog/turn-47-context-window-limits/</guid>
      <pubDate>Thu, 21 May 2026 00:00:00 GMT</pubDate>
      <description><![CDATA[Long agent sessions outgrow the context window. Why naive truncation drops task context, and how Orqen uses fill-ratio gating, hot/warm/cold history, and summarization.]]></description>
    </item>
    <item>
      <title><![CDATA[Context Caching: The LLM Cost Lever Most Agents Skip]]></title>
      <link>https://orqen.app/blog/context-caching-llm-costs/</link>
      <guid isPermaLink="true">https://orqen.app/blog/context-caching-llm-costs/</guid>
      <pubDate>Thu, 21 May 2026 00:00:00 GMT</pubDate>
      <description><![CDATA[Provider prompt caching can cut repeated-context costs up to 90%. How it works, why agents skip it, and what Orqen does when your requests aren't caching yet.]]></description>
    </item>
    <item>
      <title><![CDATA[Why 'Now Update It' Breaks Tool Routing]]></title>
      <link>https://orqen.app/blog/now-update-it-tool-routing/</link>
      <guid isPermaLink="true">https://orqen.app/blog/now-update-it-tool-routing/</guid>
      <pubDate>Wed, 20 May 2026 00:00:00 GMT</pubDate>
      <description><![CDATA[Routing on the last message alone misroutes follow-ups like 'now update it.' How Orqen uses multi-turn context and session hints — without treating every turn like a fresh chat.]]></description>
    </item>
    <item>
      <title><![CDATA[The 50KB JSON Your Agent Sends Every Turn]]></title>
      <link>https://orqen.app/blog/50kb-json-tool-result-bloat/</link>
      <guid isPermaLink="true">https://orqen.app/blog/50kb-json-tool-result-bloat/</guid>
      <pubDate>Wed, 20 May 2026 00:00:00 GMT</pubDate>
      <description><![CDATA[Bulky tool results dominate agent token cost. How Orqen shrinks role:tool messages with query-aware extraction and structural fallbacks.]]></description>
    </item>
    <item>
      <title><![CDATA[Embeddings Aren't Enough for Agent Tool Selection]]></title>
      <link>https://orqen.app/blog/embeddings-second-pass-tool-routing/</link>
      <guid isPermaLink="true">https://orqen.app/blog/embeddings-second-pass-tool-routing/</guid>
      <pubDate>Tue, 19 May 2026 00:00:00 GMT</pubDate>
      <description><![CDATA[Cosine similarity misses indirect tool dependencies. Why a second routing pass helps on chained workflows — and why it must fail open when unavailable.]]></description>
    </item>
    <item>
      <title><![CDATA[Your Tool Descriptions Are the Bug]]></title>
      <link>https://orqen.app/blog/tool-descriptions-routing-checklist/</link>
      <guid isPermaLink="true">https://orqen.app/blog/tool-descriptions-routing-checklist/</guid>
      <pubDate>Tue, 19 May 2026 00:00:00 GMT</pubDate>
      <description><![CDATA[Vague tool schemas cause routing failures before the model runs. A routing quality checklist, Orqen's schema audit, x-orqen-examples, and the dashboard view — what you fix on your side.]]></description>
    </item>
    <item>
      <title><![CDATA[MCP Gave Your Agent 50 Tools — Now What?]]></title>
      <link>https://orqen.app/blog/mcp-50-tools-now-what/</link>
      <guid isPermaLink="true">https://orqen.app/blog/mcp-50-tools-now-what/</guid>
      <pubDate>Tue, 19 May 2026 00:00:00 GMT</pubDate>
      <description><![CDATA[Connecting MCP servers is easy; stopping your agent from sending every tool on every LLM call is not. Here's why tool sprawl happens, what it costs, and how to route each turn to a small relevant subset.]]></description>
    </item>
    <item>
      <title><![CDATA[Stop Hardcoding GPT-4o: Task-Aware Model Routing]]></title>
      <link>https://orqen.app/blog/task-aware-model-routing-orqen-auto/</link>
      <guid isPermaLink="true">https://orqen.app/blog/task-aware-model-routing-orqen-auto/</guid>
      <pubDate>Mon, 18 May 2026 00:00:00 GMT</pubDate>
      <description><![CDATA[One expensive model for every turn wastes money on lookups and underpowers hard tasks. Use orqen/auto and siblings to match model capability to each request.]]></description>
    </item>
    <item>
      <title><![CDATA[What Agent Optimization Should Log (No Prompts)]]></title>
      <link>https://orqen.app/blog/agent-optimization-logging-without-prompts/</link>
      <guid isPermaLink="true">https://orqen.app/blog/agent-optimization-logging-without-prompts/</guid>
      <pubDate>Mon, 18 May 2026 00:00:00 GMT</pubDate>
      <description><![CDATA[Debug savings, recall, and latency without storing prompts. Orqen logs structured metadata per request — counts, timings, plan decisions, and honest savings math.]]></description>
    </item>
    <item>
      <title><![CDATA[Introducing Orqen: Cut Your LLM Bill Without Changing Your Code]]></title>
      <link>https://orqen.app/blog/introducing-orqen/</link>
      <guid isPermaLink="true">https://orqen.app/blog/introducing-orqen/</guid>
      <pubDate>Sun, 17 May 2026 00:00:00 GMT</pubDate>
      <description><![CDATA[Orqen sits between your agent and the LLM provider, removes the tokens the model doesn't need, and sends a smaller request — so you pay less on every call.]]></description>
    </item>
  </channel>
</rss>