<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:dc="http://purl.org/dc/elements/1.1/">
  <channel>
    <title>Telli.sh 블로그</title>
    <description>Telli.sh 팀의 소식, 제품 업데이트, AI 노트 인사이트를 전합니다.</description>
    <link>https://telli.sh/ko/blog</link>
    <atom:link href="https://telli.sh/ko/blog/rss.xml" rel="self" type="application/rss+xml"/>
    <atom:link href="https://telli.sh/blog/rss.xml" rel="alternate" type="application/rss+xml" hreflang="en" title="Telli.sh Blog"/>
    <atom:link href="https://telli.sh/ko/blog/rss.xml" rel="alternate" type="application/rss+xml" hreflang="ko" title="Telli.sh 블로그"/>
    <atom:link href="https://telli.sh/ja/blog/rss.xml" rel="alternate" type="application/rss+xml" hreflang="ja" title="Telli.sh ブログ"/>
    <atom:link href="https://telli.sh/zh/blog/rss.xml" rel="alternate" type="application/rss+xml" hreflang="zh" title="Telli.sh 博客"/>
    <atom:link href="https://telli.sh/de/blog/rss.xml" rel="alternate" type="application/rss+xml" hreflang="de" title="Telli.sh Blog"/>
    <atom:link href="https://telli.sh/fr/blog/rss.xml" rel="alternate" type="application/rss+xml" hreflang="fr" title="Telli.sh Blog"/>
    <atom:link href="https://telli.sh/es/blog/rss.xml" rel="alternate" type="application/rss+xml" hreflang="es" title="Telli.sh Blog"/>
    <atom:link href="https://telli.sh/it/blog/rss.xml" rel="alternate" type="application/rss+xml" hreflang="it" title="Telli.sh Blog"/>
    <atom:link href="https://telli.sh/pt/blog/rss.xml" rel="alternate" type="application/rss+xml" hreflang="pt" title="Telli.sh Blog"/>
    <atom:link href="https://telli.sh/ru/blog/rss.xml" rel="alternate" type="application/rss+xml" hreflang="ru" title="Telli.sh Блог"/>
    <atom:link href="https://telli.sh/pl/blog/rss.xml" rel="alternate" type="application/rss+xml" hreflang="pl" title="Telli.sh Blog"/>
    <atom:link href="https://telli.sh/vi/blog/rss.xml" rel="alternate" type="application/rss+xml" hreflang="vi" title="Telli.sh Blog"/>
    <atom:link href="https://telli.sh/th/blog/rss.xml" rel="alternate" type="application/rss+xml" hreflang="th" title="Telli.sh บล็อก"/>
    <atom:link href="https://telli.sh/id/blog/rss.xml" rel="alternate" type="application/rss+xml" hreflang="id" title="Telli.sh Blog"/>
    <atom:link href="https://telli.sh/ms/blog/rss.xml" rel="alternate" type="application/rss+xml" hreflang="ms" title="Telli.sh Blog"/>
    <atom:link href="https://telli.sh/blog/rss.xml" rel="alternate" type="application/rss+xml" hreflang="x-default" title="Telli.sh Blog"/>
    <language>ko</language>
    <lastBuildDate>Tue, 15 Sep 2026 13:17:23 GMT</lastBuildDate>
    <item>
      <title><![CDATA[구글 RSI 루머로 보는 AGI: AI 리더들은 왜 속도 조절을 말할까]]></title>
      <description><![CDATA[AGI와 재귀적 자기개선(RSI)을 쉽게 풀고 구글 루머의 X 원문을 추적합니다. 아모데이·올트먼·머스크의 속도 조절 지지, 트럼프의 중국 경쟁론, 더 빨라질 발전의 가능성과 필요한 통제 장치를 살펴봅니다.]]></description>
      <link>https://telli.sh/blog/google-rsi-rumor-agi-frontier-ko</link>
      <guid isPermaLink="true">https://telli.sh/blog/google-rsi-rumor-agi-frontier-ko</guid>
      <pubDate>Mon, 14 Sep 2026 00:00:00 GMT</pubDate>
      <dc:creator>Ken Jo</dc:creator>
      <category>agi</category>
      <category>recursive-self-improvement</category>
      <category>rsi</category>
      <category>google-deepmind</category>
      <category>ai-safety</category>
      <category>ai-policy</category>
      <category>future-of-work</category>
    </item>
    <item>
      <title><![CDATA[AI가 똑똑해질수록 지침도 손봐야 한다: 메모리·지침·프롬프트 관리법]]></title>
      <description><![CDATA[메모리, 프로젝트 규칙, 오늘 입력한 프롬프트는 모두 AI의 행동에 영향을 줍니다. 모델이 지침을 더 충실하게 따를수록 오래된 임시방편과 모호한 권한이 반복되는 문제로 굳어질 수 있습니다. AI가 기억하는 내용을 점검하고, 유용한 자율성의 범위를 정하고, 실제로 검증할 수 있는 지침을 작성하는 방법을 두 가지 템플릿과 함께 소개합니다.]]></description>
      <link>https://telli.sh/blog/ai-memory-instructions-prompts-ko</link>
      <guid isPermaLink="true">https://telli.sh/blog/ai-memory-instructions-prompts-ko</guid>
      <pubDate>Fri, 11 Sep 2026 00:00:00 GMT</pubDate>
      <dc:creator>Ken Jo</dc:creator>
      <category>ai-memory</category>
      <category>context-engineering</category>
      <category>prompting</category>
      <category>custom-instructions</category>
      <category>ai-agents</category>
      <category>evaluation</category>
    </item>
    <item>
      <title><![CDATA[출시 뒤에도 AI는 더 나아질 수 있다: 다섯 가지 개선 방식과 AGI의 가능성]]></title>
      <description><![CDATA[모델의 학습 과정만으로 능력을 모두 설명할 수는 없습니다. Reflexion과 Voyager부터 AlphaEvolve, 추론 시점 학습, 2026년의 지속 학습 실험까지, AI가 유용한 경험을 남기는 방법이 연구되고 있습니다. 실제로 무엇이 바뀌는지, 어떤 결과에 근거가 있는지, 끝없는 반복보다 믿을 수 있는 개선이 왜 중요한지 살펴봅니다.]]></description>
      <link>https://telli.sh/blog/continual-learning-ai-improvement-loops-ko</link>
      <guid isPermaLink="true">https://telli.sh/blog/continual-learning-ai-improvement-loops-ko</guid>
      <pubDate>Fri, 11 Sep 2026 00:00:00 GMT</pubDate>
      <dc:creator>Ken Jo</dc:creator>
      <category>continual-learning</category>
      <category>self-improving-ai</category>
      <category>test-time-training</category>
      <category>ai-agents</category>
      <category>agi</category>
      <category>ai-evaluation</category>
      <category>ai-memory</category>
    </item>
    <item>
      <title><![CDATA[구글의 실시간 번역, OpenAI의 전이중 음성: 언어장벽이 낮아진 뒤에 남는 일]]></title>
      <description><![CDATA[구글이 70개 언어 실시간 음성 번역을 발표한 지 3개월 만에 GPT-Live-1이 음성 1분당 $0.05로 API에 공개됐습니다. 공개된 가격과 벤치마크가 무엇을 측정하고 무엇을 측정하지 않는지, 그리고 대화가 쉬워질수록 회의록·아이디어 공유·저장한 출처가 더 중요해지는 이유를 살펴봅니다.]]></description>
      <link>https://telli.sh/blog/full-duplex-voice-language-barriers-ko</link>
      <guid isPermaLink="true">https://telli.sh/blog/full-duplex-voice-language-barriers-ko</guid>
      <pubDate>Fri, 11 Sep 2026 00:00:00 GMT</pubDate>
      <dc:creator>Ken Jo</dc:creator>
      <category>full-duplex-voice</category>
      <category>gpt-live</category>
      <category>gemini-live</category>
      <category>speech-translation</category>
      <category>meeting-notes</category>
      <category>knowledge-management</category>
    </item>
    <item>
      <title><![CDATA[사흘 동안 프런티어 모델이 넷, 그런데 어느 게 필요했는지는 아무도 말해주지 않는다]]></title>
      <description><![CDATA[2026년 9월 1일부터 3일 사이에 Anthropic, Meta, Google, OpenAI가 모두 프런티어 모델을 냈고, CNBC는 여기서 생긴 탈진에 이름을 붙였습니다. 모델 피로감입니다. 출시 횟수는 지루한 쪽입니다. 흥미로운 쪽은 출시 하나하나가 아래쪽 누군가는 반드시 돌려야 할 평가를 만들어낸다는 것, 그리고 모델을 내놓는 회사들에게는 그 평가를 건너뛰게 도와줄 이유가 전혀 없다는 것입니다. 모델 교체가 실제로 얼마를 요구하는지, 토큰당 싼 것이 왜 작업당 싼 것이 아닌지, 좁은 일 하나만을 위해 만들어진 완전 관리형 소프트웨어가 왜 남이 대신 결정해 줄 수 있는 유일한 형태인지를 짚습니다.]]></description>
      <link>https://telli.sh/blog/model-fatigue-managed-software-ko</link>
      <guid isPermaLink="true">https://telli.sh/blog/model-fatigue-managed-software-ko</guid>
      <pubDate>Tue, 08 Sep 2026 00:00:00 GMT</pubDate>
      <dc:creator>Ken Jo</dc:creator>
      <category>model-fatigue</category>
      <category>llm-evaluation</category>
      <category>ai-costs</category>
      <category>managed-software</category>
      <category>vertical-ai</category>
      <category>model-routing</category>
      <category>enterprise-ai</category>
    </item>
    <item>
      <title><![CDATA[NVIDIA는 129억 3,000만 달러로 모델 허브를 산 게 아니다. 기본값을 샀다]]></title>
      <description><![CDATA[2026년 9월 3일, NVIDIA가 Hugging Face를 12,930,300,000달러에 인수하기로 합의했습니다. Groq 자산에 쓴 200억 달러에 이은 역대 두 번째 규모입니다. Hugging Face는 2016년 십 대를 위한 챗봇 앱으로 시작해 결국 1,800만 명의 개발자에게 300만 개의 모델을 공급하는 곳이 됐습니다. 여기까지 어떻게 왔는지, NVIDIA의 지난 두 번의 플랫폼 베팅이 왜 정반대로 끝났는지, 그리고 중립성 약속이 실제로 판가름 나는 자리가 어디인지를 짚습니다. 블로그 글이 아니라 커널과 양자화 포맷, 배포 기본 대상에서 결정됩니다.]]></description>
      <link>https://telli.sh/blog/nvidia-hugging-face-acquisition-ko</link>
      <guid isPermaLink="true">https://telli.sh/blog/nvidia-hugging-face-acquisition-ko</guid>
      <pubDate>Tue, 08 Sep 2026 00:00:00 GMT</pubDate>
      <dc:creator>Ken Jo</dc:creator>
      <category>nvidia</category>
      <category>hugging-face</category>
      <category>open-source-ai</category>
      <category>open-weights</category>
      <category>acquisition</category>
      <category>ai-infrastructure</category>
      <category>transformers</category>
      <category>antitrust</category>
    </item>
    <item>
      <title><![CDATA[Clip은 웹 페이지를 요약까지 끝난 노트로 만들어 준다. 오늘부터 모든 Telli.sh 계정에 열린다.]]></title>
      <description><![CDATA[Clip은 웹 페이지를 저장만 하는 게 아니라, AI가 쓴 제목과 한 줄 요약이 붙은 Telli.sh 노트로 만들어 줍니다. 이 기능은 관리자 플래그 뒤에 가려진 채 출시됐고, 그래서 두 개의 분석 카운터는 단 한 번도 올라가지 않았습니다. 이제 그 관문이 사라졌습니다. 무료 계정도 하루 저장 30건과 AI 요약 10건을 쓰고, 브라우저 확장 프로그램 1.18 버전은 클립 저장 외에는 아무것도 할 수 없는 토큰으로 인증하는 우클릭 Save to Telli.sh를 더합니다.]]></description>
      <link>https://telli.sh/blog/clip-save-any-web-page-to-your-notes-ko</link>
      <guid isPermaLink="true">https://telli.sh/blog/clip-save-any-web-page-to-your-notes-ko</guid>
      <pubDate>Sat, 05 Sep 2026 00:00:00 GMT</pubDate>
      <dc:creator>Ken Jo</dc:creator>
      <category>clip</category>
      <category>web-clipper</category>
      <category>browser-extension</category>
      <category>product-update</category>
      <category>api-tokens</category>
      <category>privacy</category>
    </item>
    <item>
      <title><![CDATA[75분이 40분으로: GPT-6 Astra가 진짜로 잘하는 것]]></title>
      <description><![CDATA[OpenAI가 2026년 9월 3일 GPT-6 Astra를 공개하며 세계에서 가장 지능적인 모델이라고 불렀습니다. 그런데 자사 벤치마크 표는 더 날카로운 이야기를 합니다. 좁은 과제에서는 엄청난 도약(Terminal-Bench Science 22.4%→64.6%, SRE-Bench 55.9%→88.0%, OSWorld 과제당 시간 47% 단축)이 있었지만, 중립 종합 지표는 0.3점 움직였습니다. 어디가 진짜이고 어디에 각주가 붙는지, 그리고 100만 토큰 텍스트 모델이 여전히 못 하는 일을 정리했습니다.]]></description>
      <link>https://telli.sh/blog/gpt-6-astra-what-it-is-good-at-ko</link>
      <guid isPermaLink="true">https://telli.sh/blog/gpt-6-astra-what-it-is-good-at-ko</guid>
      <pubDate>Sat, 05 Sep 2026 00:00:00 GMT</pubDate>
      <dc:creator>Ken Jo</dc:creator>
      <category>gpt-6</category>
      <category>astra</category>
      <category>openai</category>
      <category>computer-use</category>
      <category>long-context</category>
      <category>ai-benchmarks</category>
      <category>agents</category>
      <category>ai-models</category>
    </item>
    <item>
      <title><![CDATA[구글 UCP는 결제 요청마다 헤더 네 개를 싣는다. 그중 셋은 나중의 분쟁을 위한 것이다]]></title>
      <description><![CDATA[Universal Commerce Protocol이 2026년 1월 11일 구글, 쇼피파이, 엣시, 웨이페어, 타깃, 월마트와 20곳이 넘는 파트너의 지지 속에 공개됐습니다. 그런데 명세를 읽어보면 쇼핑은 가장 재미없는 부분입니다. /.well-known/ucp 디스커버리, RFC 9421 메시지 서명, 웹훅 서명 의무화, 그리고 에이전트의 구매를 부인할 수 없게 만드는 것이 유일한 목적인 AP2 맨데이트 확장. UCP가 실제로 표준화한 것, 그리고 그중 구매와 아무 상관 없는 하나를 뜯어봅니다.]]></description>
      <link>https://telli.sh/blog/universal-commerce-protocol-proof-of-agreement-ko</link>
      <guid isPermaLink="true">https://telli.sh/blog/universal-commerce-protocol-proof-of-agreement-ko</guid>
      <pubDate>Sat, 05 Sep 2026 00:00:00 GMT</pubDate>
      <dc:creator>Ken Jo</dc:creator>
      <category>ucp</category>
      <category>universal-commerce-protocol</category>
      <category>agentic-commerce</category>
      <category>ai-agents</category>
      <category>ap2</category>
      <category>mcp</category>
      <category>protocols</category>
      <category>google</category>
    </item>
    <item>
      <title><![CDATA[확장 프로그램 팝업은 엉뚱한 질문에 답하고 있었다. 1.16은 대신 당신이 보고 있는 페이지를 그린다.]]></title>
      <description><![CDATA[Telli.sh: Every Translator 1.16은 툴바 팝업을 상태 화면으로 다시 만들었습니다. 지금 보고 있는 페이지의 축소 모형, 다섯 개의 라벨 칩, 그리고 확장 프로그램이 '실제로 돌아가고 있다'고 확인할 수 있을 때만 초록으로 켜지는 점. 모든 설정은 살아 있는 스켈레톤 미리보기가 있는 전용 설정 페이지로 옮겼고, 즉석 번역은 30일 뒤 스스로 정리되는 기록과 함께 자체 탭을 얻었으며, 알림 종은 이제 브라우저 언어에 맞춰 15개 로케일의 Telli.sh 블로그를 실어 나릅니다.]]></description>
      <link>https://telli.sh/blog/every-translator-extension-1-16-status-popup-ko</link>
      <guid isPermaLink="true">https://telli.sh/blog/every-translator-extension-1-16-status-popup-ko</guid>
      <pubDate>Fri, 04 Sep 2026 00:00:00 GMT</pubDate>
      <dc:creator>Ken Jo</dc:creator>
      <category>every-translator</category>
      <category>chrome-extension</category>
      <category>browser-extension</category>
      <category>translation</category>
      <category>product-update</category>
      <category>ux</category>
    </item>
    <item>
      <title><![CDATA[설치 명령어 한 줄과 봇 토큰 하나: 에이전트를 처음 돌려보는 사람을 위한 OpenClaw 2.0 세팅]]></title>
      <description><![CDATA[초보자를 위한 검증된 OpenClaw 2.0 설치 가이드입니다. 필요한 Node 버전, 에이전트에 모델을 붙이는 세 가지 방법과 각각의 비용, v2026.8.1에서 새로 들어온 6단계 가이드 온보딩이 묻는 순서, 텔레그램 4단계 페어링, 첫 주에 돌려볼 만한 작업 세 가지, 폭발 반경을 좁게 유지하는 첫날 권한 설정, 그리고 새 설치가 실제로 부딪히는 다섯 가지 실패를 모두 담았습니다.]]></description>
      <link>https://telli.sh/blog/openclaw-2-0-beginner-setup-guide-ko</link>
      <guid isPermaLink="true">https://telli.sh/blog/openclaw-2-0-beginner-setup-guide-ko</guid>
      <pubDate>Tue, 01 Sep 2026 00:00:00 GMT</pubDate>
      <dc:creator>Ken Jo</dc:creator>
      <category>openclaw</category>
      <category>ai-agents</category>
      <category>setup-guide</category>
      <category>self-hosted</category>
      <category>telegram</category>
      <category>agent-security</category>
      <category>local-llm</category>
      <category>beginners</category>
    </item>
    <item>
      <title><![CDATA[OpenClaw 2.0은 프로젝트가 여태 머지한 PR의 절반을 한 번에 냈습니다. 7주 동안 아무것도 내지 않은 뒤에요.]]></title>
      <description><![CDATA[2026년 8월 31일, 오픈소스 AI 에이전트가 v2026.8.1을 공개했습니다. 933명이 올린 16,000개 이상의 PR, 프로젝트 역사상 머지된 전체 PR의 약 절반. 230일 동안 106번 릴리스하던 팀이 49일간 의도적으로 침묵한 결과였습니다. 이제 설치 마법사가 이미 결제 중인 ChatGPT나 Claude 구독을 그대로 가져오고, 로컬 Ollama 모델도 찾아냅니다. 세션은 SQLite로 옮겨갔고 다운그레이드는 더 이상 공짜가 아닙니다. '내가 소유하는 AI'가 실제로 어떤 모양인지, 깨지는 변경점과 여전히 꺼져 있는 기본값까지 함께 짚었습니다.]]></description>
      <link>https://telli.sh/blog/openclaw-2-0-user-owned-ai-agent-ko</link>
      <guid isPermaLink="true">https://telli.sh/blog/openclaw-2-0-user-owned-ai-agent-ko</guid>
      <pubDate>Tue, 01 Sep 2026 00:00:00 GMT</pubDate>
      <dc:creator>Ken Jo</dc:creator>
      <category>openclaw</category>
      <category>ai-agents</category>
      <category>open-source</category>
      <category>local-llm</category>
      <category>agent-security</category>
      <category>multiplayer-ai</category>
      <category>bring-your-own-model</category>
      <category>ai-tooling</category>
    </item>
    <item>
      <title><![CDATA[연속 통역은 회의를 두 배로 늘린다. 실시간 번역기는 회의를 잊어버린다. 2026년 태국 팀이 실제로 저울질하는 선택지.]]></title>
      <description><![CDATA[방콕 에이전시들은 연속 통역을 시간당 3,000바트부터, 동시통역을 5,000바트부터 공시하고, 연속 통역은 회의가 굴러가는 시간을 대략 두 배로 늘립니다. 구글 번역의 Live translate는 태국어를 포함해 74개 언어를, 구글 Meet의 번역 자막은 태국어를 포함해 69개 언어를 지원합니다. 반면 Meet의 Transcripts와 Gemini 노트 작성 기능은 각각 8개 언어뿐이고, 태국어는 어느 쪽에도 없습니다. Deepgram의 Nova-3는 태국어를 th와 th-TH로 전사하면서도 10개 언어짜리 코드 스위칭 모드에서는 제외하는데, 하필 그것이 당신의 회의를 이루는 태국어–영어 혼용 문장입니다. 태국 팀을 위한 회의 번역 선택지 총정리.]]></description>
      <link>https://telli.sh/blog/thai-meeting-translation-options-ko</link>
      <guid isPermaLink="true">https://telli.sh/blog/thai-meeting-translation-options-ko</guid>
      <pubDate>Tue, 01 Sep 2026 00:00:00 GMT</pubDate>
      <dc:creator>Ken Jo</dc:creator>
      <category>thai</category>
      <category>meeting-translation</category>
      <category>interpretation</category>
      <category>live-translation</category>
      <category>google-meet</category>
      <category>meeting-notes</category>
      <category>sea</category>
      <category>asr</category>
    </item>
    <item>
      <title><![CDATA[인터뷰 하나를 담지 못하는 무료 120분: 인도네시아어 무료 AI 전사가 실제로 주는 것]]></title>
      <description><![CDATA[Notta의 무료 요금제는 한 달 전사 120분을 주고, 녹음 한 건은 3분에서 끊습니다. TurboScribe는 하루 3개 파일에 30분 상한을 겁니다. 구글 API는 무료 60분을 주고 인터페이스는 주지 않습니다. 2026년 8월 7일 Deepgram은 개선된 인도네시아어 모델을 내놨는데, 배치 경로에만 적용됐습니다. 인도네시아의 기자, 연구자, HR 면접관이 실제로 부딪히는 무료 요금제를 직접 확인해 비교했고, 어느 쪽도 광고하지 않는 한계까지 함께 적었습니다.]]></description>
      <link>https://telli.sh/blog/free-ai-interview-transcription-indonesian-ko</link>
      <guid isPermaLink="true">https://telli.sh/blog/free-ai-interview-transcription-indonesian-ko</guid>
      <pubDate>Mon, 31 Aug 2026 00:00:00 GMT</pubDate>
      <dc:creator>Ken Jo</dc:creator>
      <category>indonesian</category>
      <category>interview-transcription</category>
      <category>free-tier</category>
      <category>bahasa-indonesia</category>
      <category>asr</category>
      <category>speech-to-text</category>
      <category>sea</category>
      <category>code-switching</category>
    </item>
    <item>
      <title><![CDATA[같은 10.0% 오류율, 정반대의 결정: 회의 전사 정확도를 측정하는 법]]></title>
      <description><![CDATA[단어 오류율은 '아니다'라는 단어와 '그'라는 관사에 똑같은 무게를 매깁니다. 두 전사본이 나란히 10.0%를 받고도 그중 하나만 쓸 만한 이유가 여기 있습니다. 회의 사양에 맞춘 12분짜리 테스트 클립, 정규화 규칙을 고정한 기준 전사본, cpWER와 DER 명령어, 그리고 원시 WER에 100점 중 10점만 주는 5개 항목 채점표까지 측정 절차 전체를 정리했습니다.]]></description>
      <link>https://telli.sh/blog/measure-meeting-transcription-accuracy-ko</link>
      <guid isPermaLink="true">https://telli.sh/blog/measure-meeting-transcription-accuracy-ko</guid>
      <pubDate>Mon, 31 Aug 2026 00:00:00 GMT</pubDate>
      <dc:creator>Ken Jo</dc:creator>
      <category>transcription-accuracy</category>
      <category>word-error-rate</category>
      <category>wer</category>
      <category>cpwer</category>
      <category>diarization</category>
      <category>meeting-notes</category>
      <category>speech-to-text</category>
      <category>evaluation</category>
    </item>
    <item>
      <title><![CDATA[40개가 넘는 언어를 실시간으로 번역하고, 통화가 끝나면 지웁니다: 회의를 남기는 층은 다국어 전사입니다]]></title>
      <description><![CDATA[DeepL은 2026년 4월 16일 40개가 넘는 언어를 지원하는 Voice-to-Voice를 출시했고, 같은 제품의 FAQ에는 회의 데이터가 '메모리에서 일시적으로 처리되고 통화가 끝나면 삭제된다'고 적혀 있습니다. 8월 28일에는 Open ASR 리더보드가 첫 글로벌 사우스 언어를 추가하면서, WER 4.9로 사실상 동률인 두 모델이 화자의 출신 지역에 따라 정확도가 흔들리는 폭에서는 거의 네 배 차이가 난다는 것을 보여 줬습니다. 실시간 회의 번역은 이제 흔한 기능이 되어 갑니다. 기록은 아직 아닙니다.]]></description>
      <link>https://telli.sh/blog/multilingual-transcription-real-time-meeting-translation-ko</link>
      <guid isPermaLink="true">https://telli.sh/blog/multilingual-transcription-real-time-meeting-translation-ko</guid>
      <pubDate>Mon, 31 Aug 2026 00:00:00 GMT</pubDate>
      <dc:creator>Ken Jo</dc:creator>
      <category>multilingual-transcription</category>
      <category>real-time-meeting-translation</category>
      <category>meeting-translator</category>
      <category>deepl</category>
      <category>live-translation</category>
      <category>meeting-notes</category>
      <category>asr</category>
      <category>multilingual-meetings</category>
    </item>
    <item>
      <title><![CDATA[이름 없는 모델이 나흘 만에 26조 토큰을 태웠습니다. 가격표는 나중에 붙었죠. 24센트.]]></title>
      <description><![CDATA[2026년 8월 20일, 소속도 모델 카드도 가격도 없는 모델이 OpenRouter에 stealth/ox-alpha라는 이름으로 등장했습니다. 나흘 뒤 OpenCode 사용자들이 이 모델로 26조 토큰을 처리했고, 엿새 뒤 Z.AI가 정체를 밝혔습니다. GLM-5.3-Flash, 총 파라미터 320B, MIT 라이선스, 입력 100만 토큰당 0.15달러. 왜 요즘 연구소들은 자기 이름을 떼고 모델을 내놓는지, 그 트래픽 수치가 실제로 무엇을 측정한 것인지, 그리고 블렌디드 24센트가 음성 파이프라인 경제학에 무엇을 하는지 정리했습니다.]]></description>
      <link>https://telli.sh/blog/ox-alpha-stealth-model-glm-ko</link>
      <guid isPermaLink="true">https://telli.sh/blog/ox-alpha-stealth-model-glm-ko</guid>
      <pubDate>Mon, 31 Aug 2026 00:00:00 GMT</pubDate>
      <dc:creator>Ken Jo</dc:creator>
      <category>ox-alpha</category>
      <category>glm</category>
      <category>z-ai</category>
      <category>zhipu</category>
      <category>stealth-model</category>
      <category>openrouter</category>
      <category>open-weight</category>
      <category>ai-pricing</category>
      <category>ai-models</category>
    </item>
    <item>
      <title><![CDATA[Nova-3가 이번 달 58번째 언어를 추가했습니다. 그중 한 문장 안에서 섞이는 건 10개뿐입니다.]]></title>
      <description><![CDATA[2026년 8월 4일 Deepgram은 Nova-3에 펀자브어와 네팔어를 추가했고, 8월 7일에는 아르메니아어를 추가하면서 타밀어, 인도네시아어, 벨라루스어 모델을 개선했습니다. 9개월 전 이 모델이 다루던 언어는 31개였는데, 오늘 문서에 적힌 숫자는 58개입니다. 하지만 코드 스위칭이 되는 언어는 정확히 10개뿐이고, 8월 체인지로그 자체가 인도네시아어는 배치에서만, 벨라루스어는 스트리밍에서만 개선됐다고 적고 있습니다. 음성 인식 언어 커버리지가 늘어나는 속도와, 「지원」이라는 단어가 실제로 감추고 있는 것을 들여다봅니다.]]></description>
      <link>https://telli.sh/blog/speech-to-text-language-expansion-2026-ko</link>
      <guid isPermaLink="true">https://telli.sh/blog/speech-to-text-language-expansion-2026-ko</guid>
      <pubDate>Mon, 31 Aug 2026 00:00:00 GMT</pubDate>
      <dc:creator>Ken Jo</dc:creator>
      <category>speech-to-text</category>
      <category>deepgram</category>
      <category>nova-3</category>
      <category>multilingual</category>
      <category>language-support</category>
      <category>meeting-transcription</category>
      <category>asr</category>
      <category>code-switching</category>
    </item>
    <item>
      <title><![CDATA[구글은 당신의 베트남어 회의를 실시간으로 통역한다. 하지만 단 한 글자도 기록하지 않는다.]]></title>
      <description><![CDATA[구글 Meet의 번역 자막은 베트남어를 포함해 69개 언어를, 구글 번역의 Live translate는 74개 언어를 지원합니다. 반면 Meet의 Transcripts 기능과 Gemini 기반 노트 작성 기능은 각각 정확히 8개 언어만 지원하고, 베트남어는 어느 쪽에도 없습니다. DeepL Voice가 베트남어 음성을 받아쓸 수 있는 경로는 관리자가 따로 켜야 하고 자동 감지도 쓸 수 없는 서드파티 제공자뿐입니다. 2026년의 실시간 음성 번역이 베트남어 팀에게 실제로 주는 것과, 끝내 주지 않는 것을 정리했습니다.]]></description>
      <link>https://telli.sh/blog/vietnamese-live-voice-translation-meeting-record-ko</link>
      <guid isPermaLink="true">https://telli.sh/blog/vietnamese-live-voice-translation-meeting-record-ko</guid>
      <pubDate>Mon, 31 Aug 2026 00:00:00 GMT</pubDate>
      <dc:creator>Ken Jo</dc:creator>
      <category>vietnamese</category>
      <category>live-translation</category>
      <category>google-translate</category>
      <category>google-meet</category>
      <category>meeting-notes</category>
      <category>sea</category>
      <category>asr</category>
      <category>multilingual-meetings</category>
    </item>
    <item>
      <title><![CDATA[16.5GB짜리 파일이 2월의 최강 비공개 모델을 앞질렀습니다]]></title>
      <description><![CDATA[3년 전, 다운로드할 수 있는 최고의 모델은 HumanEval에서 GPT-4에 37점 뒤졌고 그래픽카드 두 장이 있어야 돌아갔습니다. 지난주 27B 오픈 웨이트 모델이 Artificial Analysis 인텔리전스 지수에서 52.0을 기록했습니다. Claude Opus 4.6 Max는 44.9였습니다. 가격은 9분의 1, 파일 크기는 16.5GB. 벤치마크 표와 가격 곡선, 그리고 추격 시차가 162일까지 줄어든 것의 의미를 정리했습니다.]]></description>
      <link>https://telli.sh/blog/qwen-3-8-before-after-local-ai-ko</link>
      <guid isPermaLink="true">https://telli.sh/blog/qwen-3-8-before-after-local-ai-ko</guid>
      <pubDate>Fri, 21 Aug 2026 00:00:00 GMT</pubDate>
      <dc:creator>Ken Jo</dc:creator>
      <category>open-llm</category>
      <category>qwen</category>
      <category>qwen3-8</category>
      <category>open-weight</category>
      <category>local-llm</category>
      <category>ai-pricing</category>
      <category>benchmarks</category>
      <category>ai-models</category>
    </item>
    <item>
      <title><![CDATA[5단계, 4개의 산출물: Telli.sh는 회의 녹음으로 정확히 무엇을 하는가]]></title>
      <description><![CDATA[Telli.sh 파이프라인 전체를 단계별로 살펴봅니다. 실시간 녹음과 오디오 업로드, 화자 구분이 포함된 AI 회의 전사, 44개 언어 대상 실시간 회의 번역, 10가지 포맷의 AI 요약, 그리고 공유 가능한 노트까지. 정직한 한계와 정확한 가격, 그리고 회의 전사 정확도가 실제로 무엇을 요구하는지도 함께 다룹니다.]]></description>
      <link>https://telli.sh/blog/what-tellish-does-with-a-meeting-ko</link>
      <guid isPermaLink="true">https://telli.sh/blog/what-tellish-does-with-a-meeting-ko</guid>
      <pubDate>Fri, 21 Aug 2026 00:00:00 GMT</pubDate>
      <dc:creator>Ken Jo</dc:creator>
      <category>product-update</category>
      <category>ai-meeting-transcription</category>
      <category>meeting-transcription-accuracy</category>
      <category>speaker-identification</category>
      <category>real-time-meeting-translation</category>
      <category>multilingual-meeting-notes</category>
      <category>meeting-minutes-automation</category>
      <category>ai-meeting-notes</category>
    </item>
    <item>
      <title><![CDATA[아무도 9월이라고 말하지 않았다: AI 회의 요약에 꼭 필요한 9분 검토]]></title>
      <description><![CDATA[회의록 요약을 다룬 연구에서, 실제 시의회 회의 전사본을 요약한 결과의 30.4%가 최소 한 개의 사실 오류를 포함했고 그 뒤에는 일곱 가지 오류 유형이 있었습니다. AI 회의 노트를 전사본과 대조해 검토하는 실전 가이드 — 무엇을 어떤 순서로 확인할지, 그리고 검토를 30분이 아니라 9분에 끝내려면 회의실에서 무슨 말을 해야 하는지.]]></description>
      <link>https://telli.sh/blog/ai-meeting-summary-review-ko</link>
      <guid isPermaLink="true">https://telli.sh/blog/ai-meeting-summary-review-ko</guid>
      <pubDate>Tue, 18 Aug 2026 00:00:00 GMT</pubDate>
      <dc:creator>Ken Jo</dc:creator>
      <category>meeting-notes</category>
      <category>ai-notetaker</category>
      <category>ai-summary</category>
      <category>meeting-minutes</category>
      <category>note-taking</category>
      <category>hallucination</category>
      <category>transcription</category>
    </item>
    <item>
      <title><![CDATA[번역 이어버드 네 종류, 질문은 하나: 회의가 끝나면 무엇이 남는가]]></title>
      <description><![CDATA[AirPods 라이브 번역은 10개 언어를 아이폰에서 전부 처리합니다. 449달러짜리 Timekettle W4 Pro는 52개 언어를 내겁니다. Pixel Buds는 40개지만 인터넷이 필요합니다. 2026년 8월 18일, 네 제조사의 지원 문서를 직접 읽고 마케팅이 건너뛰는 질문에 답했습니다. 대화가 끝난 뒤, 그 대화를 아직 갖고 있는 제품은 어느 것일까요?]]></description>
      <link>https://telli.sh/blog/translator-earbuds-where-does-the-record-go-ko</link>
      <guid isPermaLink="true">https://telli.sh/blog/translator-earbuds-where-does-the-record-go-ko</guid>
      <pubDate>Tue, 18 Aug 2026 00:00:00 GMT</pubDate>
      <dc:creator>Ken Jo</dc:creator>
      <category>translator-earbuds</category>
      <category>live-translation</category>
      <category>airpods</category>
      <category>pixel-buds</category>
      <category>galaxy-buds</category>
      <category>timekettle</category>
      <category>multilingual-meetings</category>
      <category>meeting-records</category>
    </item>
    <item>
      <title><![CDATA[앤트로픽이 클로드의 글에 워터마크를 넣기 시작했다 — 그리고 1,922건의 판결이 보여주는 한계]]></title>
      <description><![CDATA[2026년 8월 11일, 앤트로픽은 클로드가 쓴 글에 나중에 추적할 수 있는 표식을 넣기 시작했습니다. 이 표식은 복사·붙여넣기는 물론 번역까지 견디지만, 다시 쓰면 사라집니다. 텍스트 워터마크가 어떻게 작동하는지, 탐지 결과가 무엇을 증명하고 무엇을 증명하지 못하는지, 왜 오탐 누명의 역사가 중요한지, 그리고 왜 지금까지 기록된 1,922건의 환각 사례가 전부 이 검사를 통과했을지를 쉬운 말로 정리했습니다.]]></description>
      <link>https://telli.sh/blog/llm-text-watermarks-what-they-prove-ko</link>
      <guid isPermaLink="true">https://telli.sh/blog/llm-text-watermarks-what-they-prove-ko</guid>
      <pubDate>Sun, 16 Aug 2026 00:00:00 GMT</pubDate>
      <dc:creator>Ken Jo</dc:creator>
      <category>ai-watermark</category>
      <category>synthid</category>
      <category>ai-detection</category>
      <category>eu-ai-act</category>
      <category>ai-transparency</category>
      <category>ai-hallucination</category>
      <category>provenance</category>
      <category>ai-policy</category>
    </item>
    <item>
      <title><![CDATA[Qwen3.8-27B 오픈 웨이트: 용량, Q4_K_M, 4090·맥 실측 속도]]></title>
      <description><![CDATA[Qwen3.8-27B는 Apache 2.0입니다. Q4_K_M GGUF는 17.1GB로 32GB 노트북에 들어가고, FP8은 30.9GB입니다. 실측은 RTX 4090에서 초당 48토큰, 맥 미니 M4 Pro에서 12.75토큰.]]></description>
      <link>https://telli.sh/blog/qwen-3-8-open-weights-ko</link>
      <guid isPermaLink="true">https://telli.sh/blog/qwen-3-8-open-weights-ko</guid>
      <pubDate>Sun, 16 Aug 2026 00:00:00 GMT</pubDate>
      <dc:creator>Ken Jo</dc:creator>
      <category>open-llm</category>
      <category>qwen</category>
      <category>qwen3-8</category>
      <category>open-weight</category>
      <category>local-llm</category>
      <category>quantization</category>
      <category>gguf</category>
      <category>ai-models</category>
    </item>
    <item>
      <title><![CDATA[음성 인식도 조용히 오픈 웨이트가 됐다 — 용량은 1.56GB]]></title>
      <description><![CDATA[알리바바 Qwen3-ASR 모델은 아파치 2.0 라이선스에 52개 언어와 방언을 지원하고, 내려받으면 2기가바이트가 안 됩니다. 오픈 웨이트 전사 계층을 쉬운 말로 정리했습니다. 지금 무엇이 나와 있는지, 내 노트북에서 직접 돌리면 실제로 무엇을 얻는지, 그리고 닫힌 API가 여전히 이기는 지점은 어디인지.]]></description>
      <link>https://telli.sh/blog/open-weight-speech-recognition-ko</link>
      <guid isPermaLink="true">https://telli.sh/blog/open-weight-speech-recognition-ko</guid>
      <pubDate>Wed, 05 Aug 2026 00:00:00 GMT</pubDate>
      <dc:creator>Ken Jo</dc:creator>
      <category>speech-to-text</category>
      <category>open-weight</category>
      <category>qwen3-asr</category>
      <category>whisper</category>
      <category>parakeet</category>
      <category>transcription</category>
      <category>self-hosting</category>
      <category>ai-models</category>
    </item>
    <item>
      <title><![CDATA[Pro는 $8, Pro+는 $10: 더 단순해지고 더 싸진 Telli.sh 가격]]></title>
      <description><![CDATA[7월 20일부터 Telli.sh 유료 플랜 가격이 다시 내려갔습니다. Pro는 월 $8, Pro+는 월 $10, 연간 플랜은 15% 할인을 그대로 유지합니다. 변경 전후 비교표와 가격을 둥글게 다듬어 내린 이유를 정리했습니다.]]></description>
      <link>https://telli.sh/blog/tellish-simpler-lower-pricing-ko</link>
      <guid isPermaLink="true">https://telli.sh/blog/tellish-simpler-lower-pricing-ko</guid>
      <pubDate>Wed, 05 Aug 2026 00:00:00 GMT</pubDate>
      <dc:creator>Ken Jo</dc:creator>
      <category>pricing</category>
      <category>product-update</category>
      <category>ai-notes</category>
      <category>meeting-notes</category>
      <category>transcription</category>
      <category>accessibility</category>
    </item>
    <item>
      <title><![CDATA[중국의 3주 오픈 웨이트 질주: Kimi K3, Qwen3.8-Max, 그리고 이 흐름을 끝낼 수도 있는 반전]]></title>
      <description><![CDATA[3주 사이 중국의 대표 AI 연구소들이 프론티어급 오픈 웨이트 모델을 공개하거나 예고했습니다 — Kimi K3, Qwen3.8-Max, 그리고 정식 출시에 이른 DeepSeek V4. 각 모델이 무엇인지, 무엇이 실제로 공개됐고 무엇이 아직 약속인지, 그리고 베이징이 왜 자신이 키운 흐름을 이제 규제하려 하는지를 쉬운 말로 풀어봅니다.]]></description>
      <link>https://telli.sh/blog/china-open-weight-model-wave-ko</link>
      <guid isPermaLink="true">https://telli.sh/blog/china-open-weight-model-wave-ko</guid>
      <pubDate>Tue, 04 Aug 2026 00:00:00 GMT</pubDate>
      <dc:creator>Ken Jo</dc:creator>
      <category>open-llm</category>
      <category>kimi-k3</category>
      <category>qwen</category>
      <category>deepseek</category>
      <category>open-weight</category>
      <category>china-ai</category>
      <category>ai-models</category>
      <category>meeting-notes</category>
    </item>
    <item>
      <title><![CDATA[재회의(再會議): 같은 결정을 두 번 하지 않으려면]]></title>
      <description><![CDATA[2026년 7월 OpenAI 연구에 따르면 직무 특이적 ChatGPT 메시지의 43.5%가 자기 직무 바깥의 일이었습니다. 그런 일은 맥락을 먹고 자라는데, 조직의 맥락 대부분은 회의에서 증발합니다. 회의를 검색 가능한 업무 지식으로 바꾸는 실전 가이드 — 무엇을 남기고, 어떻게 이름 붙이고, 어떤 회의를 기록으로 남길지, 그리고 검색이 온보딩과 인수인계를 어떻게 바꾸는지.]]></description>
      <link>https://telli.sh/blog/meetings-to-searchable-knowledge-ko</link>
      <guid isPermaLink="true">https://telli.sh/blog/meetings-to-searchable-knowledge-ko</guid>
      <pubDate>Tue, 04 Aug 2026 00:00:00 GMT</pubDate>
      <dc:creator>Ken Jo</dc:creator>
      <category>meeting-notes</category>
      <category>knowledge-management</category>
      <category>searchable-transcripts</category>
      <category>team-knowledge</category>
      <category>onboarding</category>
      <category>ai-at-work</category>
      <category>productivity</category>
    </item>
    <item>
      <title><![CDATA[회의는 실패하지 않았습니다. 그다음 24시간이 실패했습니다]]></title>
      <description><![CDATA[병목은 회의가 아니라 그 다음 날입니다. 결정은 흐릿해지고, 담당자는 모호해지고, 후속 작업은 DM 스레드에서 사라집니다. 실행하는 팀이 실제로 돌리는 노트 루프를 짚어 봅니다. 결정·담당자·기한을 회의 중에 적고, 몇 분 안에 공유하고, 다음 회의를 그 노트로 여는 방법입니다.]]></description>
      <link>https://telli.sh/blog/notes-for-teams-that-execute-ko</link>
      <guid isPermaLink="true">https://telli.sh/blog/notes-for-teams-that-execute-ko</guid>
      <pubDate>Tue, 04 Aug 2026 00:00:00 GMT</pubDate>
      <dc:creator>Ken Jo</dc:creator>
      <category>meeting-notes</category>
      <category>action-items</category>
      <category>meeting-follow-up</category>
      <category>team-productivity</category>
      <category>decision-tracking</category>
      <category>remote-work</category>
      <category>note-taking</category>
    </item>
    <item>
      <title><![CDATA[단어 오류율 2.4%, 그런데 누가 말했는지는 모른다]]></title>
      <description><![CDATA[마이크로소프트 MAI-Transcribe-1.5는 단어 오류율 2.4%를 기록했습니다. 그리고 같은 모델의 공식 문서에는 '화자 분리는 지원하지 않습니다'라고 적혀 있습니다. WER이 무엇을 측정하는지, 왜 리더보드가 바닥에 몰려 있는지, 그리고 전사를 쓸 만한 회의 노트로 바꾸는 것은 결국 음성 인식 위의 어떤 층인지 쉽게 정리했습니다.]]></description>
      <link>https://telli.sh/blog/stt-accuracy-is-not-enough-ko</link>
      <guid isPermaLink="true">https://telli.sh/blog/stt-accuracy-is-not-enough-ko</guid>
      <pubDate>Tue, 04 Aug 2026 00:00:00 GMT</pubDate>
      <dc:creator>Ken Jo</dc:creator>
      <category>speech-to-text</category>
      <category>word-error-rate</category>
      <category>transcription</category>
      <category>diarization</category>
      <category>meeting-notes</category>
      <category>mai-transcribe</category>
      <category>ai-notetaker</category>
    </item>
    <item>
      <title><![CDATA[전사는 이제 AI가 씁니다 — 그래도 좋은 회의 노트에 반드시 남아야 하는 여섯 가지]]></title>
      <description><![CDATA[Granola는 AI 노트를 손목에 올렸고, VOMO는 사용자 40만 명을 넘겼으며, 오픈소스 대안은 몇 주 만에 등장했습니다. 그중 어떤 것도 '좋은 회의 노트'의 조건을 바꾸지는 못했습니다. 노트에 반드시 남아야 하는 여섯 가지와, 60초 자가 점검법.]]></description>
      <link>https://telli.sh/blog/what-good-meeting-notes-keep-ai-era-ko</link>
      <guid isPermaLink="true">https://telli.sh/blog/what-good-meeting-notes-keep-ai-era-ko</guid>
      <pubDate>Tue, 04 Aug 2026 00:00:00 GMT</pubDate>
      <dc:creator>Ken Jo</dc:creator>
      <category>meeting-notes</category>
      <category>note-taking</category>
      <category>ai-notetaker</category>
      <category>meeting-minutes</category>
      <category>productivity</category>
      <category>ai-transcription</category>
    </item>
    <item>
      <title><![CDATA[2026년 중반 오픈 웨이트의 급부상: Kimi K3, GLM-5.2, 그리고 오픈 모델이 계속 좋아지는 이유]]></title>
      <description><![CDATA[Kimi K3는 2026년 오픈 웨이트 흐름의 주인공이지만, 지금 바로 내려받아 쓸 수 있는 모델은 GLM-5.2입니다. 두 모델을 솔직하게 정리하고, 관대한 라이선스의 오픈 모델이 회의 노트에 왜 중요한지 살펴봅니다.]]></description>
      <link>https://telli.sh/blog/open-llm-kimi-k3-glm-52-ko</link>
      <guid isPermaLink="true">https://telli.sh/blog/open-llm-kimi-k3-glm-52-ko</guid>
      <pubDate>Mon, 20 Jul 2026 00:00:00 GMT</pubDate>
      <dc:creator>Ken Jo</dc:creator>
      <category>open-llm</category>
      <category>kimi-k3</category>
      <category>glm-5.2</category>
      <category>open-weight</category>
      <category>ai-models</category>
      <category>meeting-notes</category>
      <category>ai-notetaker</category>
    </item>
    <item>
      <title><![CDATA[GPT-5.6 Sol·Terra·Luna: 모델 등급과 추론 강도]]></title>
      <description><![CDATA[OpenAI는 gpt-5.6 별칭을 Sol로 보내고, 균형은 Terra, 대량 처리는 Luna로 나눴습니다. 추론 강도는 none부터 max까지이며 기본값은 medium — GPT-5.5에서 옮길 때 무엇이 달라지는지 정리했습니다.]]></description>
      <link>https://telli.sh/blog/gpt-5p6-launch-meeting-intelligence-ko</link>
      <guid isPermaLink="true">https://telli.sh/blog/gpt-5p6-launch-meeting-intelligence-ko</guid>
      <pubDate>Sat, 11 Jul 2026 00:00:00 GMT</pubDate>
      <dc:creator>Ken Jo</dc:creator>
      <category>gpt-5.6</category>
      <category>ai-models</category>
      <category>meeting-intelligence</category>
      <category>meeting-notes</category>
      <category>transcription</category>
      <category>ai-notetaker</category>
    </item>
    <item>
      <title><![CDATA[Fable 5와 Mythos 5: AI 모델 접근권도 제품 신뢰의 일부가 되었습니다]]></title>
      <description><![CDATA[Anthropic의 Fable 5 접근 중단과 복구 사례는 안전한 AI 모델에도 명확한 보호장치, 원본 기록, 설명 가능한 사용자 제어가 필요하다는 점을 보여줍니다.]]></description>
      <link>https://telli.sh/blog/fable-5-ai-governance-safe-model-access-ko</link>
      <guid isPermaLink="true">https://telli.sh/blog/fable-5-ai-governance-safe-model-access-ko</guid>
      <pubDate>Fri, 03 Jul 2026 00:00:00 GMT</pubDate>
      <dc:creator>Ken Jo</dc:creator>
      <category>ai-governance</category>
      <category>anthropic</category>
      <category>fable-5</category>
      <category>mythos-5</category>
      <category>frontier-models</category>
      <category>cybersecurity</category>
    </item>
    <item>
      <title><![CDATA[Deepgram STT: 화자 분리 v2, 숫자 표기, 배치와 실시간의 차이]]></title>
      <description><![CDATA[Deepgram의 화자 분리 v2는 기존 프로덕션 분리기보다 3.3배 선호됐습니다. 다만 diarize_model은 배치 전용이라 스트리밍에는 쓸 수 없습니다. 한국어 띄어쓰기와 다국어 숫자 표기도 함께 개선됐습니다.]]></description>
      <link>https://telli.sh/blog/deepgram-stt-updates-live-ai-notes-ko</link>
      <guid isPermaLink="true">https://telli.sh/blog/deepgram-stt-updates-live-ai-notes-ko</guid>
      <pubDate>Sun, 28 Jun 2026 00:00:00 GMT</pubDate>
      <dc:creator>Ken Jo</dc:creator>
      <category>deepgram</category>
      <category>speech-to-text</category>
      <category>live-notes</category>
      <category>meeting-notes</category>
      <category>transcription</category>
      <category>ai-notetaker</category>
    </item>
    <item>
      <title><![CDATA[GPT-5.6와 GPT-5.5: 2026년 6월 OpenAI 공식 문서 기준]]></title>
      <description><![CDATA[날짜가 붙은 스냅숏입니다. 2026년 6월 28일 기준 OpenAI 최신 모델 안내는 GPT-5.6이 아니라 여전히 GPT-5.5를 가리켰습니다. 추론, 도구 사용, 롱컨텍스트에 대해 무엇이라 적혀 있었는지 정리했습니다.]]></description>
      <link>https://telli.sh/blog/gpt-5p6-gpt-5p5-meeting-notes-system-of-record-ko</link>
      <guid isPermaLink="true">https://telli.sh/blog/gpt-5p6-gpt-5p5-meeting-notes-system-of-record-ko</guid>
      <pubDate>Sun, 28 Jun 2026 00:00:00 GMT</pubDate>
      <dc:creator>Ken Jo</dc:creator>
      <category>gpt-5.5</category>
      <category>gpt-5.6</category>
      <category>ai-models</category>
      <category>meeting-notes</category>
      <category>transcription</category>
      <category>ai-notetaker</category>
    </item>
    <item>
      <title><![CDATA[Telli.sh 유료 플랜 가격을 낮췄습니다]]></title>
      <description><![CDATA[더 많은 사용자가 라이브 AI 노트, 전사, 요약, 다국어 회의 흐름의 혜택을 누릴 수 있도록 Telli.sh 유료 플랜 가격을 낮췄습니다.]]></description>
      <link>https://telli.sh/blog/tellish-lower-pricing-announcement-ko</link>
      <guid isPermaLink="true">https://telli.sh/blog/tellish-lower-pricing-announcement-ko</guid>
      <pubDate>Sun, 28 Jun 2026 00:00:00 GMT</pubDate>
      <dc:creator>Ken Jo</dc:creator>
      <category>pricing</category>
      <category>product-update</category>
      <category>ai-notes</category>
      <category>meeting-notes</category>
      <category>transcription</category>
      <category>accessibility</category>
    </item>
    <item>
      <title><![CDATA[Google 실시간 음성 번역이 보여주는 것: 다국어 회의에는 검토 가능한 노트가 필요합니다]]></title>
      <description><![CDATA[Google은 Meet과 Translate에 실시간 음성 번역을 확장하고 있습니다. Telli.sh는 번역 순간 이후에도 남아야 하는 검토 가능한 회의 기록에 집중합니다.]]></description>
      <link>https://telli.sh/blog/google-live-speech-translation-reviewable-notes-ko</link>
      <guid isPermaLink="true">https://telli.sh/blog/google-live-speech-translation-reviewable-notes-ko</guid>
      <pubDate>Thu, 25 Jun 2026 00:00:00 GMT</pubDate>
      <dc:creator>Ken Jo</dc:creator>
      <category>live-translation</category>
      <category>google-meet</category>
      <category>gemini</category>
      <category>ai-notes</category>
      <category>meeting-notes</category>
      <category>multilingual-meetings</category>
    </item>
    <item>
      <title><![CDATA[경찰 AI 전사 실험이 보여주는 것: 중요한 대화에는 검토 가능한 기록이 필요합니다]]></title>
      <description><![CDATA[South Australia Police는 인터뷰 전사를 위한 AI speech-to-text 실험을 준비하고 있습니다. 핵심은 속도가 아니라 검토 가능한 원문 기록입니다.]]></description>
      <link>https://telli.sh/blog/police-ai-transcription-reviewable-records-ko</link>
      <guid isPermaLink="true">https://telli.sh/blog/police-ai-transcription-reviewable-records-ko</guid>
      <pubDate>Thu, 25 Jun 2026 00:00:00 GMT</pubDate>
      <dc:creator>Ken Jo</dc:creator>
      <category>ai-transcription</category>
      <category>speech-to-text</category>
      <category>public-sector-ai</category>
      <category>interview-transcription</category>
      <category>ai-notes</category>
      <category>reviewable-records</category>
    </item>
    <item>
      <title><![CDATA[Telli.sh: Every Translator를 Chrome 웹 스토어에서 사용할 수 있습니다]]></title>
      <description><![CDATA[무료 Google 번역 모드, Chrome 내장 온디바이스 AI, 원문 옆 번역 보기를 지원하는 Telli.sh: Every Translator가 Chrome 웹 스토어에 출시되었습니다.]]></description>
      <link>https://telli.sh/blog/every-translator-chrome-web-store-ko</link>
      <guid isPermaLink="true">https://telli.sh/blog/every-translator-chrome-web-store-ko</guid>
      <pubDate>Tue, 23 Jun 2026 00:00:00 GMT</pubDate>
      <dc:creator>Ken Jo</dc:creator>
      <category>every-translator</category>
      <category>chrome-extension</category>
      <category>browser-extension</category>
      <category>translation</category>
      <category>product-update</category>
    </item>
  </channel>
</rss>