<?xml version="1.0" encoding="UTF-8"?>
<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9"
        xmlns:news="http://www.google.com/schemas/sitemap-news/0.9">
  <url>
    <loc>https://samtaelabs.com/blog/esp32-s3-llm-cluster-edge-ai-decision-criteria</loc>
    <news:news>
      <news:publication>
        <news:name>삼태연구소</news:name>
        <news:language>ko</news:language>
      </news:publication>
      <news:publication_date>2026-10-01T12:30:36.324Z</news:publication_date>
      <news:title>ESP32-S3 클러스터로 LLM을 돌릴 수 있다는 것과, 그게 우리 제품에 맞는 선택인지는 다른 문제다</news:title>
    </news:news>
  </url>
  <url>
    <loc>https://samtaelabs.com/blog/scissor-vs-illustrator-figma-design-tool-adoption-criteria</loc>
    <news:news>
      <news:publication>
        <news:name>삼태연구소</news:name>
        <news:language>ko</news:language>
      </news:publication>
      <news:publication_date>2026-10-01T03:00:28.353Z</news:publication_date>
      <news:title>Scissor, Illustrator·Figma의 진짜 대안인가: 파일 호환성·협업·보안 기준으로 나눠보는 도입 범위</news:title>
    </news:news>
  </url>
  <url>
    <loc>https://samtaelabs.com/blog/ai-agent-guardrail-model-based-vs-deterministic-data-flow</loc>
    <news:news>
      <news:publication>
        <news:name>삼태연구소</news:name>
        <news:language>ko</news:language>
      </news:publication>
      <news:publication_date>2026-09-30T22:30:03.361Z</news:publication_date>
      <news:title>AI 에이전트 가드레일, 모델 기반 승인과 결정론적 데이터 흐름 통제를 어떤 기준으로 조합할까</news:title>
    </news:news>
  </url>
  <url>
    <loc>https://samtaelabs.com/blog/browser-micro-llm-vs-cloud-llm-task-routing-decision-guide</loc>
    <news:news>
      <news:publication>
        <news:name>삼태연구소</news:name>
        <news:language>ko</news:language>
      </news:publication>
      <news:publication_date>2026-09-30T12:30:36.478Z</news:publication_date>
      <news:title>브라우저 내 초소형 LLM, 어떤 기능에 쓰고 어떤 작업은 클라우드에 남겨야 하는가</news:title>
    </news:news>
  </url>
  <url>
    <loc>https://samtaelabs.com/blog/field-inspection-app-photo-report-action-tracking-scope</loc>
    <news:news>
      <news:publication>
        <news:name>삼태연구소</news:name>
        <news:language>ko</news:language>
      </news:publication>
      <news:publication_date>2026-09-30T03:00:28.345Z</news:publication_date>
      <news:title>현장 점검 앱, 사진 보고서로 끝낼지 조치 완료까지 관리할지 정하는 법</news:title>
    </news:news>
  </url>
  <url>
    <loc>https://samtaelabs.com/blog/ai-agent-hook-recursive-loop-cost-control-design</loc>
    <news:news>
      <news:publication>
        <news:name>삼태연구소</news:name>
        <news:language>ko</news:language>
      </news:publication>
      <news:publication_date>2026-09-29T22:30:03.289Z</news:publication_date>
      <news:title>에이전트 훅이 자기 자신을 다시 부를 때: 재귀 실행과 비용 폭주를 막는 실행 경계 설계</news:title>
    </news:news>
  </url>
</urlset>