<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom">
  <channel>
    <title>BlueMouse.ai Insights</title>
    <link>https://www.bluemouse.ai/insights</link>
    <atom:link href="https://www.bluemouse.ai/insights/feed.xml" rel="self" type="application/rss+xml"/>
    <description>Practical articles on building RAG, OCR and document AI systems.</description>
    <language>en-gb</language>
    <item>
      <title>Why legal RAG needs hybrid search: reciprocal rank fusion, explained with lawyers&#x27; queries</title>
      <link>https://www.bluemouse.ai/insights/legal-rag-hybrid-search-rrf</link>
      <guid isPermaLink="true">https://www.bluemouse.ai/insights/legal-rag-hybrid-search-rrf</guid>
      <pubDate>Mon, 14 Sep 2026 09:00:00 +0000</pubDate>
      <category>Legal RAG</category>
      <description>Lawyers mix exact citations with plain-language questions. Why legal RAG needs hybrid retrieval, how reciprocal rank fusion merges results, and how to test it.</description>
    </item>
    <item>
      <title>Invoice OCR is the easy part. Validation is what lets finance trust the result.</title>
      <link>https://www.bluemouse.ai/insights/invoice-ocr-validation-before-automation</link>
      <guid isPermaLink="true">https://www.bluemouse.ai/insights/invoice-ocr-validation-before-automation</guid>
      <pubDate>Thu, 17 Sep 2026 09:00:00 +0000</pubDate>
      <category>OCR and document processing</category>
      <description>Reading an invoice is the easy part of invoice OCR. A worked example shows the checks that turn extracted fields into data finance can post, and what happens when one fails.</description>
    </item>
  </channel>
</rss>
