<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Observer on Medium]]></title>
        <description><![CDATA[Stories by Observer on Medium]]></description>
        <link>https://medium.com/@observerfoundation?source=rss-504fed9cb408------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*12FmGU6_Sb9qv4_W3JldFg.png</url>
            <title>Stories by Observer on Medium</title>
            <link>https://medium.com/@observerfoundation?source=rss-504fed9cb408------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Sun, 22 Nov 2026 11:18:09 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@observerfoundation/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[옵저버, 하이브리드 강수 진단 연구 논문 출간]]></title>
            <link>https://medium.com/obsr/%EC%98%B5%EC%A0%80%EB%B2%84-%ED%95%98%EC%9D%B4%EB%B8%8C%EB%A6%AC%EB%93%9C-%EA%B0%95%EC%88%98-%EC%A7%84%EB%8B%A8-%EC%97%B0%EA%B5%AC-%EB%85%BC%EB%AC%B8-%EC%B6%9C%EA%B0%84-70cab046e3ae?source=rss-504fed9cb408------2</link>
            <guid isPermaLink="false">https://medium.com/p/70cab046e3ae</guid>
            <category><![CDATA[notice]]></category>
            <dc:creator><![CDATA[Observer]]></dc:creator>
            <pubDate>Mon, 09 Nov 2026 03:21:06 GMT</pubDate>
            <atom:updated>2026-11-09T03:24:43.884Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*pgQEjViWWv0EmtXFV41OPQ.png" /></figure><p>옵저버가 집필한 하이브리드 강수 진단 AI ‘FuXi-OPE’ 관련 연구 논문이 ‘한국지구과학회지(Korean Earth Science Society Journal) AI 연구 특별호’에 정식 채택되어 출간되었습니다.</p><p>기존의 딥러닝 예보 모델들은 극한 기상이나 국지성 호우 발생 시 강수 강도가 낮아지는 ‘과소모의(Underestimation Bias)’ 한계를 지니고 있었습니다.</p><p>이를 보완하기 위해 옵저버는 딥러닝 예보 모델로 거시적인 종관 역학을 예단(Prognostic)하고,</p><p>국지적 강수는 3차원 연직 상태를 해석하는 진단적(Diagnostic) 후처리(FuXi-OPE)를 결합한 하이브리드 전략을 적용했습니다.</p><p>본 기술 개발을 통해 집중호우의 강도와 광역 강수대 예측 성능이 개선되었습니다.</p><p>특히 중기 예보(3~6일 후) 구간에서 비구름의 위치 일치도 및 강한 비가 내리는 지점을 정확히 맞춘 비율(CSI)과 실제 내린 비의 양과 예측한 양의 비율(빈도 편향)이 크게 향상되었습니다.</p><p>‘FuXi-OPE’의 중기 예보 구간에서 국지성 호우 탐지 및 유지력 개선 효과는 공신력 있는 학계 검증을 거쳐 객관적으로 입증되었습니다.</p><p>이번 결과는 옵저버 AI 기상 예측 기술이 다음 단계로 도약하고 있음을 보여주는 성과입니다.</p><p>앞으로도 실제 예측 결과에 대한 검증과 기술 개선을 지속하여, 실효성 높은 기상 정보를 제공하겠습니다.</p><p>옵저버 AI의 기술 개선과 향후 행보에 많은 관심 부탁드립니다.</p><p>감사합니다.</p><p>Observer’s research paper on ‘FuXi-OPE’, a hybrid AI system for precipitation diagnosis, has been formally accepted and published in the ‘AI Research Special Issue of the Journal of the Korean Earth Science Society’.</p><p>Existing deep learning weather forecasting models have tended to underestimate rainfall intensity during extreme weather events and localized heavy downpours a limitation known as underestimation bias.</p><p>To address this limitation, Observer adopted a hybrid strategy that combines a deep learning forecasting model with diagnostic post-processing. The forecasting model predicts large-scale synoptic atmospheric dynamics through a prognostic approach, while ‘FuXi-OPE’ uses a diagnostic approach to estimate local precipitation by analyzing the three-dimensional atmospheric state and its vertical structure.</p><p>This approach has improved the prediction of heavy-rainfall intensity and large-scale precipitation bands.</p><p>In particular, at medium-range lead times of three to six days, the system showed substantial improvements in the spatial agreement between forecast and observed precipitation patterns, the Critical Success Index (CSI) for heavy rainfall detection, and frequency bias.</p><p>FuXi-OPE’s improved ability to detect and retain localized heavy-rainfall signals in medium-range forecasts has been objectively demonstrated through academic validation.</p><p>These results mark a further step forward in Observer’s AI weather prediction technology.</p><p>We will continue to validate our forecasts against actual observations and refine our technology to deliver useful and actionable weather information.</p><p>We appreciate your continued interest in Observer AI’s technological progress and future developments.</p><p>Thank you.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=70cab046e3ae" width="1" height="1" alt=""><hr><p><a href="https://medium.com/obsr/%EC%98%B5%EC%A0%80%EB%B2%84-%ED%95%98%EC%9D%B4%EB%B8%8C%EB%A6%AC%EB%93%9C-%EA%B0%95%EC%88%98-%EC%A7%84%EB%8B%A8-%EC%97%B0%EA%B5%AC-%EB%85%BC%EB%AC%B8-%EC%B6%9C%EA%B0%84-70cab046e3ae">옵저버, 하이브리드 강수 진단 연구 논문 출간</a> was originally published in <a href="https://medium.com/obsr">OBSR</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[옵저버 AI 일기예보 V5 출시]]></title>
            <link>https://medium.com/obsr/%EC%98%B5%EC%A0%80%EB%B2%84-ai-%EC%9D%BC%EA%B8%B0%EC%98%88%EB%B3%B4-v5-%EC%B6%9C%EC%8B%9C-f65ac7852234?source=rss-504fed9cb408------2</link>
            <guid isPermaLink="false">https://medium.com/p/f65ac7852234</guid>
            <category><![CDATA[development]]></category>
            <category><![CDATA[notice]]></category>
            <dc:creator><![CDATA[Observer]]></dc:creator>
            <pubDate>Tue, 21 Jul 2026 05:11:05 GMT</pubDate>
            <atom:updated>2026-07-21T05:12:01.120Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*oomDFmNfs_ZNDl8fKLdoFA.jpeg" /></figure><p>더욱 발전된 기상예측 경험을 제공하기 위해 옵저버 AI 일기예보의 새로운 예측 엔진 V5를 선보입니다.</p><p>V5는 그동안 축적한 기상 데이터와 예측 결과를 바탕으로 AI 분석 체계를 한층 고도화한 버전입니다. 변화하는 기상 상황을 정교하게 분석하고, 예측의 정확도와 안정성을 높이는 데 중점을 두었습니다.</p><p>V5는 현재 옵저버 앱에서 바로 선택하여 이용하실 수 있습니다. 기존 V1·V2·V3·V4의 예측 결과도 함께 제공되므로, 각 버전의 예보를 비교하며 기상 변화를 보다 입체적으로 확인하실 수 있습니다.</p><p>이번 V5 출시는 옵저버 AI 기상예측 기술의 또 한 번의 발전을 의미합니다. 앞으로도 실제 예측 결과에 대한 지속적인 검증과 기술 개선을 이어가며, 일상과 다양한 의사결정에 실질적으로 도움이 되는 기상 정보를 제공하겠습니다.</p><p>새로워진 옵저버 AI 일기예보 V5에 많은 관심과 이용 부탁드립니다.</p><p>감사합니다.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/865/1*ByZWFBypKJye1n2oMIJLIg.jpeg" /></figure><p>We are pleased to introduce V5, the latest prediction engine for the Observer AI Weather Forecast service, designed to deliver a more advanced weather forecasting experience.</p><p>Built on accumulated weather data and previous forecasting results, V5 features an enhanced AI analysis framework. This new version focuses on analyzing changing weather conditions with greater precision while improving overall forecasting accuracy and reliability.</p><p>V5 is now available for selection within the Observer app. Users can also compare its forecasts with those generated by V1, V2, V3, and V4, providing a more comprehensive view of changing weather conditions.</p><p>The launch of V5 marks another important step forward in the development of Observer’s AI-powered weather forecasting technology. We will continue to validate actual forecasting results and refine our technology to provide weather information that offers practical value for everyday life and informed decision-making.</p><p>We invite you to explore the newly upgraded Observer AI Weather Forecast V5.</p><p>Thank you.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=f65ac7852234" width="1" height="1" alt=""><hr><p><a href="https://medium.com/obsr/%EC%98%B5%EC%A0%80%EB%B2%84-ai-%EC%9D%BC%EA%B8%B0%EC%98%88%EB%B3%B4-v5-%EC%B6%9C%EC%8B%9C-f65ac7852234">옵저버 AI 일기예보 V5 출시</a> was originally published in <a href="https://medium.com/obsr">OBSR</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[AI기반 날씨예보서비스V4 출시]]></title>
            <link>https://medium.com/obsr/ai%EA%B8%B0%EB%B0%98-%EB%82%A0%EC%94%A8%EC%98%88%EB%B3%B4%EC%84%9C%EB%B9%84%EC%8A%A4v4-%EC%B6%9C%EC%8B%9C-95378152904f?source=rss-504fed9cb408------2</link>
            <guid isPermaLink="false">https://medium.com/p/95378152904f</guid>
            <category><![CDATA[development]]></category>
            <category><![CDATA[notice]]></category>
            <dc:creator><![CDATA[Observer]]></dc:creator>
            <pubDate>Fri, 12 Jun 2026 03:18:48 GMT</pubDate>
            <atom:updated>2026-06-12T03:19:07.824Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*atYq0nqtFsfBPdiNiA9fCA.jpeg" /></figure><p>옵저버 AI 일기예보 서비스의 새로운 예측 엔진인 V4 버전이 정식 출시되었습니다.<br>V4는 기존 버전 대비 더욱 향상된 AI 알고리즘을 적용하여 기상 변화에 대한 분석 능력과 예측 성능을 한층 강화하였습니다.<br>현재 앱 내에서 V4를 선택하여 사용할 수 있으며, V1•V2•V3와 함께 비교 분석도 가능합니다.<br>당사는 지속적인 검증과 성능 개선을 통해 보다 정확하고 신뢰할 수 있는 AI 일기예보 서비스를 제공하기 위해 노력하고 있습니다.<br>앞으로도 옵저버 AI는 인공지능 기반 기상예측 기술의 발전을 통해 사용자들에게 더욱 가치 있는 기상 정보를 제공하겠습니다.<br>감사합니다.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/865/1*ByZWFBypKJye1n2oMIJLIg.jpeg" /></figure><p>The V4 version, a new prediction engine for the Observer AI Weather Forecast service, has been officially released.</p><p>Compared to previous versions, V4 applies an enhanced AI algorithm, further strengthening its ability to analyze weather changes and improve forecasting performance.</p><p>V4 can now be selected and used within the app, and users can also compare its forecasts with V1, V2, and V3.</p><p>We will continue to carry out ongoing verification and performance improvements to provide a more accurate and reliable AI weather forecasting service.</p><p>Observer AI will continue advancing AI-based weather prediction technology to deliver more valuable weather information to users.</p><p>Thank you.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=95378152904f" width="1" height="1" alt=""><hr><p><a href="https://medium.com/obsr/ai%EA%B8%B0%EB%B0%98-%EB%82%A0%EC%94%A8%EC%98%88%EB%B3%B4%EC%84%9C%EB%B9%84%EC%8A%A4v4-%EC%B6%9C%EC%8B%9C-95378152904f">AI기반 날씨예보서비스V4 출시</a> was originally published in <a href="https://medium.com/obsr">OBSR</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[AI 날씨예보 예측시스템(V4·V5) 및 통합본 출시 계획 안내]]></title>
            <link>https://medium.com/obsr/ai-%EB%82%A0%EC%94%A8%EC%98%88%EB%B3%B4-%EC%98%88%EC%B8%A1%EC%8B%9C%EC%8A%A4%ED%85%9C-v4-v5-%EB%B0%8F-%ED%86%B5%ED%95%A9%EB%B3%B8-%EC%B6%9C%EC%8B%9C-%EA%B3%84%ED%9A%8D-%EC%95%88%EB%82%B4-1a62162cf6fc?source=rss-504fed9cb408------2</link>
            <guid isPermaLink="false">https://medium.com/p/1a62162cf6fc</guid>
            <category><![CDATA[development]]></category>
            <category><![CDATA[notice]]></category>
            <dc:creator><![CDATA[Observer]]></dc:creator>
            <pubDate>Thu, 04 Jun 2026 06:34:53 GMT</pubDate>
            <atom:updated>2026-06-04T06:38:42.549Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*aGeBp8mj_9GBQfcTVLDc0w.jpeg" /></figure><p>곧 출시 예정인 새로운 AI 날씨예보 서비스에 대해 안내드립니다.</p><p>이번 신규 서비스는 기존에 출시된 V1, V2, V3 예측시스템과는 별도로 제공되는 새로운 예측시스템입니다.</p><p>당초 V4 이후 통합본 출시를 안내드렸으나, 통합본의 예측 정확도와 완성도를 높이기 위해 추가 예측시스템인 V5까지 순차적으로 제공한 뒤 최종 통합본을 선보일 예정입니다.</p><p>V5는 단순히 버전을 늘리기 위한 것이 아니라, 보다 정교한 통합 AI 날씨예보 서비스를 위한 과정으로 이해해주시길 바랍니다. 최종 통합본에서는 각 예측시스템의 장점을 종합하여 정확한 날씨예보 정보를 제공할 수 있도록 하겠습니다.</p><p><strong>V4. Swin Transformer 기반 예측 시스템 (6월 출시 예정)</strong></p><p>V4 모델은 스윈 트랜스포머(Swin Transformer) 구조를 기상학적 특성에 맞게 확장한 ‘3차원 지구특화 트랜스포머(3D Earth-Specific Transformer, 3DEST)’를 기반으로 하는 딥러닝 일기예보 모델입니다. 본 모델은 3차원 입력 자료를 입체 큐브 단위로 패치화한 뒤, 컨볼루션 레이어를 통해 하나의 연속적인 벡터 공간으로 변환하여 인코더-디코더 트랜스포머 블록에 입력합니다. 이 과정에서 구면 투영에 따른 위치 왜곡을 방지하기 위해 지구특화 위치편향(Earth-Specific Positional Bias, ESPB)을 적용하였으며, 장기 예측 시 발생하는 오차 누적을 억제하고자 계층적 시간 통합(Hierarchical Temporal Aggregation) 방식을 도입했습니다. 기존 V1, V2 모델과 달리, 강수량을 직접 예측하는 대신 3차원 예측 필드에 적응형 푸리에 신경 연산자(Adaptive Fourier Neural Operator, AFNO)를 적용하여 간접 추정하는 방식을 채택함으로써 강수 예측의 정밀도 높였습니다.</p><p><strong>V5. Graph Neural Networks 기반 예측 시스템</strong></p><p>V5 모델은 V2 모델과 마찬가지로 그래프 신경망(Graph Neural Networks, GNN) 기반의 인코더-프로세서-디코더 구조를 활용하는 딥러닝 일기예보 모델입니다. 다만, 공간 격자 체계와 어텐션(Attention) 메커니즘에서 뚜렷한 차별성을 가집니다. 구체적으로 본 모델은 구면 왜곡을 최소화하기 위해 팔면체 감축 가우시안 격자(O96 Octahedral Grid)를 도입하였으며, 연산 효율성을 높이기 위해 트랜스포머 기반의 슬라이딩 윈도우 어텐션(Sliding-Window Attention)을 적용했습니다. 뿐만 아니라 예측의 현실성을 보장하기 위해 강수량이 음수 값으로 산출되지 않도록 하는 등 수학적·물리적 제약 조건을 모델 구조 내에 강제하였습니다.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/865/1*ByZWFBypKJye1n2oMIJLIg.jpeg" /></figure><p>We would like to provide an update regarding our upcoming new AI weather forecasting service.</p><p>This new service will be offered as a separate forecasting system from the previously released V1, V2, and V3 prediction systems.</p><p>We had initially announced that the integrated version would be released after V4. However, in order to further improve the forecast accuracy and overall completeness of the integrated version, we have decided to release an additional prediction system, V5, in sequence before launching the final integrated version.</p><p>V5 should not be understood as simply adding another version, but rather as part of the process of building a more advanced and refined integrated AI weather forecasting service.</p><p>In the final integrated version, we will combine the strengths of each prediction system to provide more accurate weather forecast information.</p><p><strong>V4. Swin Transformer-Based Forecasting System</strong><br><strong>Scheduled for release in June</strong></p><p>The V4 model is a deep learning-based weather forecasting model built on a Three-Dimensional Earth-Specific Transformer (3DEST), which extends the Swin Transformer architecture to better reflect meteorological characteristics. The model divides three-dimensional input data into volumetric cube-based patches and converts them into a continuous vector space through convolutional layers before feeding them into encoder-decoder transformer blocks.</p><p>To prevent positional distortion caused by spherical projection, the model applies Earth-Specific Positional Bias (ESPB). It also incorporates a Hierarchical Temporal Aggregation method to reduce the accumulation of errors that can occur in long-range forecasting.</p><p>Unlike the existing V1 and V2 models, V4 does not directly predict precipitation. Instead, it applies an Adaptive Fourier Neural Operator (AFNO) to the three-dimensional forecast fields to estimate precipitation indirectly, thereby improving the precision of precipitation forecasting.</p><p><strong>V5. Graph Neural Network-Based Forecasting System</strong></p><p>The V5 model is a deep learning-based weather forecasting model that, like the V2 model, uses an encoder-processor-decoder architecture based on Graph Neural Networks (GNNs). However, it differs significantly in its spatial grid system and attention mechanism.</p><p>Specifically, the model adopts an O96 Octahedral Grid to minimize spherical distortion and applies transformer-based Sliding-Window Attention to improve computational efficiency.</p><p>In addition, to ensure the physical realism of its forecasts, the model enforces mathematical and physical constraints within its architecture, such as preventing precipitation values from being calculated as negative numbers.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=1a62162cf6fc" width="1" height="1" alt=""><hr><p><a href="https://medium.com/obsr/ai-%EB%82%A0%EC%94%A8%EC%98%88%EB%B3%B4-%EC%98%88%EC%B8%A1%EC%8B%9C%EC%8A%A4%ED%85%9C-v4-v5-%EB%B0%8F-%ED%86%B5%ED%95%A9%EB%B3%B8-%EC%B6%9C%EC%8B%9C-%EA%B3%84%ED%9A%8D-%EC%95%88%EB%82%B4-1a62162cf6fc">AI 날씨예보 예측시스템(V4·V5) 및 통합본 출시 계획 안내</a> was originally published in <a href="https://medium.com/obsr">OBSR</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[옵저버 로드맵 발표(2026)]]></title>
            <link>https://medium.com/obsr/%EC%98%B5%EC%A0%80%EB%B2%84-%EB%A1%9C%EB%93%9C%EB%A7%B5-%EB%B0%9C%ED%91%9C-2026-93a54eaca3a4?source=rss-504fed9cb408------2</link>
            <guid isPermaLink="false">https://medium.com/p/93a54eaca3a4</guid>
            <category><![CDATA[notice]]></category>
            <dc:creator><![CDATA[Observer]]></dc:creator>
            <pubDate>Wed, 21 Jan 2026 06:16:07 GMT</pubDate>
            <atom:updated>2026-01-21T06:33:04.028Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*H-dssgs3vpTh_ROkdrTGiw.jpeg" /></figure><p>옵저버는 급변하는 기술 환경과 기후 리스크의 구조적 변화를 기회로 삼아, AI 기반 일기예보 기술을 핵심 성장 축으로 하는 사업 전략을 본격적으로 추진하고자 합니다. 기존의 단순 정보 제공을 넘어, 데이터 신뢰성과 예측 정확도를 지속적으로 고도화한 차세대 AI 일기예보 플랫폼을 통해 산업·금융·생활 전반에 실질적인 의사결정 가치를 제공하는 것이 올해 회사의 핵심 방향성입니다. 당사는 축적된 기술력과 검증 프로세스를 바탕으로 AI 예보 기술의 상업적 활용 가능성을 단계적으로 확대하며, 중장기적으로는 글로벌 시장에서도 경쟁력 있는 AI 기후 데이터 기업으로 도약하고자 합니다.</p><p><strong>AI 일기예보 모델 추가 개발</strong></p><p>OBSERVER AI는 일기예보 모델 V4를 개발하고 있습니다. 자세한 내용은 다음과 같습니다. V4 모델은 그래프 신경망(Graph Neural Networks) 인코더 및 디코더와 슬라이딩 윈도우 트랜스포머(Sliding Window Transformer) 프로세스 구조를 활용한 딥러닝 일기예보 모델입니다. 트랜스포머 모델이 주로 사용하는 셀프 어텐션(Self-Attention) 방식의 비효율성을 극복하기 위해, 특정 지역을 예측할 때 일정 범위(윈도우) 내의 데이터를 집중적으로 참조하고, 이 윈도우를 이동시키면서 전체를 파악하는 방식을 활용합니다.</p><p><strong>AI 기반 다중모델 앙상블 예측 시스템 개발</strong></p><p>향후 개발이 완료될 V4를 포함하여 총 4개의 AI 일기예보 모델을 활용하여 종합적인 예측 시스템을 개발할 예정입니다. 3–4개의 모델이 제공하는 기온과 강수량 예측 정보에 기반하여, 평균, 최대, 최소값을 동시에 제공함으로써 여러 예측 모델을 개별적으로 확인해야 하는 번거로움을 줄이고 보다 정확한 예측 정보를 제공하고자 합니다.</p><p><strong>연장중기 AI예보 개발</strong></p><p>옵저버는 10일 예보의 한계를 넘어, 최대 6주까지 연장중기 AI예보를 도입하고자 합니다. 예측 모델은 Transformer 기반의 인코더-디코더 구조로 설계됩니다. Swin Transformer블록을 사용하여 고차원 기상 데이터를 계층적으로 처리하고, 시간에 따라 빠르게 변하는 기상패턴을 학습하기 위해 Spatio-temporal Knowledge Bridge 모듈을 활용합니다. 이 모듈은 기상 변수간 시공간적 상관관계를 강화하여, 연장중기 예측에서 발생할 수 있는 데이터의 왜곡을 방지하고 물리적 흐름의 연속성을 유지하는 역할을 합니다. 예측 결과는 주간 평균 기온과 강수량을 중심으로 제공하고자 합니다.</p><p><strong>AI일기예보 데이터 정확성 검증</strong></p><p>AI일기예보 시스템은 다양한 기상 데이터와 알고리즘을 기반으로 예측 결과를 산출하나, AI 모델의 특성상 예측 정확도는 지속적인 데이터 검증을 통해 점진적으로 개선되는 구조를 가지고 있습니다. 이에 따라 당사는 다음과 같은 목적 하에 AI 예보 데이터의 정확성 및 신뢰성 검증 절차를 올해 진행할 계획입니다. 1) 실제 관측 데이터와 AI 예측 결과 간의 정합성 검증, 2) 예보 정확도 지표(오차율, 적중률 등)에 대한 정기적 분석, 3) 데이터 품질 개선 및 알고리즘 고도화를 위한 지속적 학습 등.</p><p><strong>AI 일기예보 시스템 상업적 활용 계획</strong></p><p>AI 기반 일기예보 시스템에 대해 데이터 정확성 및 신뢰성 검증 절차를 단계적으로 수행한 이후,해당 검증이 일정 수준 이상 충족될 경우, 본 시스템을 다양한 산업 분야에 상업적으로 활용하는 방안을 검토·추진할 계획입니다.</p><p>상업적 활용은 다음과 같은 방향을 중심으로 단계적으로 진행될 예정입니다.</p><ul><li><strong><em>기업·기관 대상 맞춤형 기상 예측 서비스</em></strong></li><li><strong><em>산업별(농업, 물류, 에너지, 보험 등) 의사결정 지원용 예보 데이터 제공</em></strong></li><li><strong><em>API 및 구독형 서비스 모델을 통한 데이터 서비스 제공</em></strong></li></ul><p>다만, 본 계획은 기술적 검증 결과, 시장 환경, 관련 법·제도 및 내부 정책 등에 따라 변경 또는 조정될 수 있으며, 상업적 활용 시점 및 범위에 대해서는 확정되지 않았습니다.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/865/1*ByZWFBypKJye1n2oMIJLIg.jpeg" /></figure><p>Observer seeks to actively advance a business strategy that positions AI-based weather forecasting technology as its core growth engine by turning rapid technological changes and structural shifts in climate risk into strategic opportunities. Moving beyond the provision of simple information, the company’s key strategic direction this year is to deliver tangible decision-making value across industry, finance, and daily life through a next-generation AI weather forecasting platform that continuously enhances data reliability and predictive accuracy. Building on its accumulated technological expertise and validation processes, the company plans to gradually expand the commercial applicability of its AI forecasting technologies and, over the mid-to-long term, to evolve into a globally competitive AI climate data company.</p><p><strong>Additional Development of AI Weather Forecasting Models</strong></p><p>OBSERVER AI is currently developing Weather Forecasting Model V4, with the following key features. The V4 model is a deep learning–based weather forecasting system that employs a Graph Neural Network (GNN) encoder–decoder architecture combined with a Sliding Window Transformer processing framework. To address the inefficiencies of the self-attention mechanism commonly used in transformer models, the system focuses on selectively referencing data within a defined window when forecasting specific regions, while shifting the window to capture the broader spatial context.</p><p><strong>Development of an AI-Based Multi-Model Ensemble Forecasting System</strong></p><p>Including the V4 model currently under development, a total of four AI weather forecasting models will be integrated into a comprehensive ensemble forecasting system. Based on temperature and precipitation forecasts generated by three to four models, the system will simultaneously provide average, maximum, and minimum values. This approach aims to reduce the need to review multiple forecasting models individually while delivering more accurate and reliable predictive information.</p><p><strong>Development of Extended Medium-Range AI Forecasting</strong></p><p>OBSERVER aims to move beyond the limitations of conventional 10-day forecasts by introducing extended medium-range AI forecasts of up to six weeks. The medium-range forecasting model is designed using a transformer-based encoder–decoder architecture. Swin Transformer blocks are employed to hierarchically process high-dimensional meteorological data, while a Spatio-temporal Knowledge Bridge module is used to learn rapidly evolving weather patterns over time. This module strengthens spatiotemporal correlations among meteorological variables, helping to prevent data distortion that may arise in extended medium-range forecasting and to maintain continuity in physical atmospheric processes. Forecast outputs will primarily focus on weekly average temperature and precipitation.</p><p><strong>Validation of AI Weather Forecast Data Accuracy</strong></p><p>AI weather forecasting systems generate predictions based on diverse meteorological data and algorithms. However, they requires continuous validation for accuracy improvement. Accordingly, OBSERVER plans to conduct verification procedures this year to assess the accuracy and reliability of AI forecast data, with the following objectives: 1) Validation of consistency between observed data and AI-generated forecasts, 2) Regular analysis of forecast accuracy metrics (e.g., error rates, hit rates), and 3) Continuous learning to improve data quality and further advance algorithms.</p><p><strong>Commercial Utilization of the AI Weather Forecasting System</strong></p><p>After completing step-by-step validation of data accuracy and reliability for the AI-based weather forecasting system, OBSERVER plans to review and pursue the commercial application of the system across various industries. Commercial utilization will be implemented in phases, focusing on the following areas: customized weather forecasting services for enterprises, provision of forecast data to support decision-making across industries (e.g., agriculture, logistics, energy, insurance), delivery of data services through API-based and subscription-based service models. Note that these plans may be subject to change or adjustment depending on technical validation outcomes, market conditions, relevant laws and regulations, and internal policies. The timing and scope of commercial utilization have not yet been finalized.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=93a54eaca3a4" width="1" height="1" alt=""><hr><p><a href="https://medium.com/obsr/%EC%98%B5%EC%A0%80%EB%B2%84-%EB%A1%9C%EB%93%9C%EB%A7%B5-%EB%B0%9C%ED%91%9C-2026-93a54eaca3a4">옵저버 로드맵 발표(2026)</a> was originally published in <a href="https://medium.com/obsr">OBSR</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[AI기반 날씨예보서비스V3 출시]]></title>
            <link>https://medium.com/obsr/ai%EA%B8%B0%EB%B0%98-%EB%82%A0%EC%94%A8%EC%98%88%EB%B3%B4%EC%84%9C%EB%B9%84%EC%8A%A4v3-%EC%B6%9C%EC%8B%9C-6cac205e0af2?source=rss-504fed9cb408------2</link>
            <guid isPermaLink="false">https://medium.com/p/6cac205e0af2</guid>
            <category><![CDATA[development]]></category>
            <category><![CDATA[notice]]></category>
            <dc:creator><![CDATA[Observer]]></dc:creator>
            <pubDate>Thu, 21 Aug 2025 01:41:01 GMT</pubDate>
            <atom:updated>2025-08-21T01:41:01.038Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*JZyFadzzEk_MSyOvCVpeHw.jpeg" /></figure><p>AI기반 날씨예보서비스 버전3가 출시되었습니다.</p><p>V3는 Cross-modal Transformer 기반 예측 시스템이며 모달 맞춤형 인코더, 크로스 모달 fusion transformer, 모달 맞춤형 디코더 구조를 활용한 딥러닝 일기예보 모델입니다.</p><p>다중 모달 네트워크 구조과 다중 작업 최적화를 위한 불확실성 로스를 활용하여 다양한 변수를 모달리티로 인식하고, 변수간 상호작용을 크로스 모달 transformer를 통해 모델링합니다. 최적 가중치를 자동으로 학습하고, 변수별 불확실성을 고려해 가중치를 조정합니다. 기온 예측에 최적화 되어 있으며, 강수는 예측하지 않습니다 (강수관련 변수는 V2를 활용합니다)</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*JrPenfoKma3vBqwOwHqbhw.png" /></figure><p>날씨예보서비스 버전3는 기존의 옵저버AI날씨앱에서 이용 가능합니다.<br>사용자들은 앱의 우측하단의 V3, V2, V1 버튼을 선택하여 기존의 예보정보와 비교하며 이용하실 수 있습니다.<br>V1~V3의 예보 정확도 비교 분석을 통해 예보서비스를 지속적으로 발전시켜 나가도록 하겠습니다.<br>감사합니다.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/865/1*ByZWFBypKJye1n2oMIJLIg.jpeg" /></figure><p>AI-based Weather Forecasting Service V3 Launched</p><p>We are pleased to announce the release of Version 3 (V3) of our AI-based weather forecasting service.</p><p>V3 is a Cross-modal Transformer-based prediction system, utilizing modality-specific encoders, a cross-modal fusion transformer, and modality-specific decoders. This deep learning model employs a multi-modal network structure and applies uncertainty loss for multi-task optimization, enabling the system to recognize various variables as modalities and to model their interactions through the cross-modal transformer. The system automatically learns the optimal weights and adjusts them by accounting for variable-specific uncertainties.</p><p>The model is optimized for temperature prediction and does not forecast precipitation (precipitation-related variables are handled by V2)</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*JrPenfoKma3vBqwOwHqbhw.png" /></figure><p>V3 is available on the existing Observer AI Weather App. Users can select the V3, V2, or V1 buttons on the bottom-right corner of the app to compare forecasts across different versions. We will continue to enhance our forecasting service through comparative accuracy analysis of V1–V3.</p><p>Thank you.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=6cac205e0af2" width="1" height="1" alt=""><hr><p><a href="https://medium.com/obsr/ai%EA%B8%B0%EB%B0%98-%EB%82%A0%EC%94%A8%EC%98%88%EB%B3%B4%EC%84%9C%EB%B9%84%EC%8A%A4v3-%EC%B6%9C%EC%8B%9C-6cac205e0af2">AI기반 날씨예보서비스V3 출시</a> was originally published in <a href="https://medium.com/obsr">OBSR</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[AI 기상예보 서비스 ‘폭염지수’ 도입]]></title>
            <link>https://medium.com/obsr/ai-%EA%B8%B0%EC%83%81%EC%98%88%EB%B3%B4-%EC%84%9C%EB%B9%84%EC%8A%A4-%ED%8F%AD%EC%97%BC%EC%A7%80%EC%88%98-%EB%8F%84%EC%9E%85-b6e845a8ba14?source=rss-504fed9cb408------2</link>
            <guid isPermaLink="false">https://medium.com/p/b6e845a8ba14</guid>
            <category><![CDATA[development]]></category>
            <category><![CDATA[notice]]></category>
            <dc:creator><![CDATA[Observer]]></dc:creator>
            <pubDate>Wed, 30 Jul 2025 00:52:31 GMT</pubDate>
            <atom:updated>2025-07-30T00:52:31.886Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*swrCvhYhkghfsLCR0Y3OWw.jpeg" /></figure><p>옵저버 AI 날씨지수에 폭염지수(Heat Index) 기능이 새롭게 추가되었습니다.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/677/1*L4azpDwwfqf_gh6RBl5QAw.png" /></figure><p>폭염지수는 단순한 기온 정보뿐 아니라 체감온도와 습도를 종합적으로 반영하여, 건강에 미치는 영향을 보다 직관적으로 알려드립니다.<br>무더위가 계속되는 여름철, 옵저버 AI기상예보와 함께 대비하시기 바랍니다.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/865/1*ByZWFBypKJye1n2oMIJLIg.jpeg" /></figure><p>Implementation of Heat Index in the Observer AI Weather Forecasting Service.<br>The Heat Index feature has been newly introduced in the Observer AI Weather Index.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/677/1*L4azpDwwfqf_gh6RBl5QAw.png" /></figure><p>This index reflects not only the actual temperature but also incorporates humidity to provide a comprehensive measure of perceived heat and its potential health impact.<br>Users are advised to refer to this information for appropriate preparation during periods of extreme heat.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=b6e845a8ba14" width="1" height="1" alt=""><hr><p><a href="https://medium.com/obsr/ai-%EA%B8%B0%EC%83%81%EC%98%88%EB%B3%B4-%EC%84%9C%EB%B9%84%EC%8A%A4-%ED%8F%AD%EC%97%BC%EC%A7%80%EC%88%98-%EB%8F%84%EC%9E%85-b6e845a8ba14">AI 기상예보 서비스 ‘폭염지수’ 도입</a> was originally published in <a href="https://medium.com/obsr">OBSR</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[AI기반 날씨예보서비스V2 출시]]></title>
            <link>https://medium.com/obsr/ai%EA%B8%B0%EB%B0%98-%EB%82%A0%EC%94%A8%EC%98%88%EB%B3%B4%EC%84%9C%EB%B9%84%EC%8A%A4v2-%EC%B6%9C%EC%8B%9C-56e6929e8af5?source=rss-504fed9cb408------2</link>
            <guid isPermaLink="false">https://medium.com/p/56e6929e8af5</guid>
            <category><![CDATA[development]]></category>
            <dc:creator><![CDATA[Observer]]></dc:creator>
            <pubDate>Mon, 16 Jun 2025 06:09:57 GMT</pubDate>
            <atom:updated>2025-06-16T06:21:56.138Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*QVJbcR7c6iNjzOKcFk5Twg.jpeg" /></figure><p>AI기반 날씨예보서비스 V2출시</p><p>AI기반 날씨예보서비스 버전2가 출시되었습니다.</p><p>V2 모델은 그래프 신경망(Graph Neural Networks) 기반의 인코더-프로세서-디코더 구조를 활용하는 딥러닝 일기예보 모델입니다 (이전 안내공지 참고)</p><p>날씨예보서비스 버전2는 기존의 옵저버AI날씨앱에서 이용 가능합니다.</p><p>사용자들은 앱의 우측하단의 V2, V1 버튼을 선택하여 기존의 예보정보와 비교하며 이용하실 수 있습니다.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/795/1*ct9ZeryPoSUuPRq-pzVqSg.png" /></figure><p>날씨예보의 정확도 개선을 위하여 예보서비스는 지속적으로 고도화해나갈 계획입니다.</p><p>감사합니다.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/865/1*ByZWFBypKJye1n2oMIJLIg.jpeg" /></figure><p>AI-Powered Weather Forecasting Service V2 Launched</p><p>We’re excited to announce the release of Version 2 of our AI-powered weather forecasting service!</p><p>The V2 model utilizes a deep learning weather forecasting model based on a Graph Neural Network (GNN) encoder-processor-decoder architecture. (Please refer to our previous announcement for more details.)</p><p>You can access Weather Forecasting Service Version 2 within the existing Observer AI Weather app.</p><p>Users can compare the new V2 forecasts with the previous ones by selecting the “V2” or “V1” buttons located at the bottom right of the app.</p><p>We plan to continuously enhance our forecasting service to improve the accuracy of weather predictions.</p><p>Thank you</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=56e6929e8af5" width="1" height="1" alt=""><hr><p><a href="https://medium.com/obsr/ai%EA%B8%B0%EB%B0%98-%EB%82%A0%EC%94%A8%EC%98%88%EB%B3%B4%EC%84%9C%EB%B9%84%EC%8A%A4v2-%EC%B6%9C%EC%8B%9C-56e6929e8af5">AI기반 날씨예보서비스V2 출시</a> was originally published in <a href="https://medium.com/obsr">OBSR</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[AI기반 신규 날씨예보서비스 안내]]></title>
            <link>https://medium.com/obsr/ai%EA%B8%B0%EB%B0%98-%EC%8B%A0%EA%B7%9C-%EB%82%A0%EC%94%A8%EC%98%88%EB%B3%B4%EC%84%9C%EB%B9%84%EC%8A%A4-%EC%95%88%EB%82%B4-1115f65c41d8?source=rss-504fed9cb408------2</link>
            <guid isPermaLink="false">https://medium.com/p/1115f65c41d8</guid>
            <category><![CDATA[development]]></category>
            <dc:creator><![CDATA[Observer]]></dc:creator>
            <pubDate>Wed, 11 Jun 2025 07:21:57 GMT</pubDate>
            <atom:updated>2025-06-16T05:49:34.510Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*YEyp0adkUdGM08PQMiEfKg.jpeg" /></figure><p>출시 예정인 새로운 AI날씨예보서비스에 대하여 안내드립니다.</p><p>기존 출시한 AI날씨예보서비스와 별도로 서비스할 예정으로, 사용자들은 두가지 다른 예보를 비교하며 선택적으로 이용하실 수 있습니다.</p><p>▶ 기존 예보모델</p><p>U-Transformer 기반 예측 시스템<br>V1 모델은 U-Net과 Transformer의 장점을 결합한 U-Transformer 아키텍처를 사용하는 딥러닝 일기예보 모델입니다. Downsampling과 Upsampling 경로를 통해 다양한 스케일의 기상 패턴을 포착하며, Self-Attention 메커니즘을 활용하여 장거리 현상과의 상호작용을 모델링 합니다. 또한 3차원 Cube Embedding을 통하여 기상 데이터의 공간적 특성을 포착합니다. Skip Connection을 통해 서로 다른 스케일의 정보를 효율적으로 통합하며, 최종적으로 Fully Connected Layer를 통해 각 기상 변수에 대한 예측이 이루어집니다.</p><p>▶ 신규 예보모델</p><p>Graph Neural Networks 기반 예측 시스템<br>V2 모델은 그래프 신경망(Graph Neural Networks) 기반의 인코더-프로세서-디코더 구조를 활용하는 딥러닝 일기예보 모델입니다. 입력 데이터는 인코더를 통해 그래프 형태로 변환되고, 프로세서에서 메시지 전달 알고리즘을 통해 노드 값을 업데이트하고, 디코더를 통해 원래의 위경도 격자로 데이터를 변환해 예측 결과를 출력합니다. 그래프는 지구 전체를 균일한 공간 해상도로 나타내는 Multimesh로 구성되며, 이를 통해 국지적인 정보 뿐만 아니라 장거리 정보까지 효율적으로 예측에 활용합니다.</p><p>신규서비스는 6월 중 출시를 목표로 하고 있습니다.<br>추가적인 내용들에 대해서는 이후 공지로 안내 드리도록 하겠습니다.</p><p>감사합니다.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/865/1*ByZWFBypKJye1n2oMIJLIg.jpeg" /></figure><p>Introducing New AI-Powered Weather Forecasting Service</p><p>We’re excited to announce an upcoming AI-powered weather forecasting service!<br>This new service will be offered separately from our existing AI weather forecast, allowing users to compare and choose between two distinct forecast model.</p><p>Current Forecasting Model</p><p>U-Transformer-Based Prediction System</p><p>V1 model is a deep learning weather forecasting model that utilizes a U-Transformer architecture, combining the strengths of U-Net and Transformer. It captures weather patterns at various scales through downsampling and upsampling paths and models interactions with long-range phenomena using a Self-Attention mechanism. Furthermore, 3D Cube Embedding is employed to capture the spatial characteristics of meteorological data. Skip Connections efficiently integrate information from different scales, and finally, predictions for each meteorological variable are made through a Fully Connected Layer.</p><p>New Forecasting Model</p><p>Graph Neural Networks-Based Prediction System</p><p>V2 model is a deep learning weather forecasting model that uses an Encoder-Processor-Decoder structure based on Graph Neural Networks. Input data is transformed into a graph representation by the encoder, node values are updated through a message-passing algorithm in the processor, and the decoder transforms the data back into the original latitude-longitude grid to release prediction results. The graph is composed of a Multimesh, representing the entire Earth with uniform spatial resolution, which efficiently utilizes both local and long-range information for predictions.</p><p>We are targeting a June launch for this new service. Further details will be provided in subsequent announcements.</p><p>Thank you.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=1115f65c41d8" width="1" height="1" alt=""><hr><p><a href="https://medium.com/obsr/ai%EA%B8%B0%EB%B0%98-%EC%8B%A0%EA%B7%9C-%EB%82%A0%EC%94%A8%EC%98%88%EB%B3%B4%EC%84%9C%EB%B9%84%EC%8A%A4-%EC%95%88%EB%82%B4-1115f65c41d8">AI기반 신규 날씨예보서비스 안내</a> was originally published in <a href="https://medium.com/obsr">OBSR</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[옵저버 로드맵 발표(2025)]]></title>
            <link>https://medium.com/obsr/%EC%98%B5%EC%A0%80%EB%B2%84-%EB%A1%9C%EB%93%9C%EB%A7%B5-%EB%B0%9C%ED%91%9C-2025-1c59692edaa5?source=rss-504fed9cb408------2</link>
            <guid isPermaLink="false">https://medium.com/p/1c59692edaa5</guid>
            <category><![CDATA[notice]]></category>
            <dc:creator><![CDATA[Observer]]></dc:creator>
            <pubDate>Tue, 31 Dec 2024 05:50:51 GMT</pubDate>
            <atom:updated>2024-12-31T05:50:51.152Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*TVQxd5NXperJcuBqjz5_UQ.jpeg" /></figure><p>2025년 계획은 아래와 같습니다. 세부적인 진행사항에 대해서는 별도의 공지를 통하여 안내드리도록 하겠습니다.</p><ul><li>해외 관측망 확대</li><li>AI 날씨 예보 고도화(앙상블, 바이어스커넥션 등)</li><li>AI 날씨예보서비스 매출처 유치 마케팅 (B2B, B2G)</li><li>AI를 활용한 날씨 연계 서비스 개발</li><li>옵저버앱 개편 (제보방식 등)</li><li>AI농업기상예보 추진(장기적 과업)</li><li>주요 파트너사 유치</li><li>데이터스토어 런칭</li></ul><p>상기 계획은 사업의 진행 상황에 따라 변경될 수 있습니다.<br>팀은 실제 가치를 창출하기 위하여 최선을 다하겠습니다.<br>2025년 많은 관심과 참여 부탁 드립니다.</p><p>감사합니다.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/715/1*37rdJzc9qdi9r6DAkZmBrw.jpeg" /></figure><p>Following are the plans for 2025. Detailed updates will be provided by separate announcements.</p><ul><li>Expansion of the overseas observation network</li><li>Advancement of AI weather forecasting (ensemble, bias correction, etc.)</li><li>Marketing to generate revenue for AI weather forecast (B2B, B2G)</li><li>Development of weather-related services using AI</li><li>Revamping of the Observer App (e.g., reporting methods)</li><li>Promotion of AI agricultural weather forecast (long-term project)</li><li>Create more partnership</li><li>Data Store launch</li></ul><p>The above plans are subject to change depending on the business.</p><p>The project team will make the best effort to create values. We look forward to your continued interest and participation in 2025.</p><p>Thank you</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=1c59692edaa5" width="1" height="1" alt=""><hr><p><a href="https://medium.com/obsr/%EC%98%B5%EC%A0%80%EB%B2%84-%EB%A1%9C%EB%93%9C%EB%A7%B5-%EB%B0%9C%ED%91%9C-2025-1c59692edaa5">옵저버 로드맵 발표(2025)</a> was originally published in <a href="https://medium.com/obsr">OBSR</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
    </channel>
</rss>