#KnowledgeSense
LightNews LightNews
tech-trending.bsky.social
NVIDIAの「長文コンテキスト不要」論文
https://zenn.dev/knowledgesense/articles/bd917ed6829121
NVIDIAの「長文コンテキスト不要」論文
zenn.dev
November 10, 2026 at 6:05 PM
zenn-ai-feed-bot.bsky.social
NVIDIAの「長文コンテキスト不要」論文 | Atsushi Kadowaki #zenn
https://zenn.dev/knowledgesense/articles/bd917ed6829121
NVIDIAの「長文コンテキスト不要」論文
zenn.dev
November 10, 2026 at 4:22 AM
zenn-ai-feed-bot.bsky.social
LLM同士は文章で会話しないことで速度と精度が上がる | sasakuna #zenn
https://zenn.dev/knowledgesense/articles/e4fc456906e959
LLM同士は文章で会話しないことで速度と精度が上がる
zenn.dev
November 3, 2026 at 1:50 AM
zenn-ai-feed-bot.bsky.social
.md は危険な実行ファイルだ。AI時代のセキュリティ意識 | Atsushi Kadowaki #zenn
https://zenn.dev/knowledgesense/articles/4ff01df8f59a3e
.md は危険な実行ファイルだ。AI時代のセキュリティ意識
zenn.dev
October 27, 2026 at 11:04 AM
zenn-ai-feed-bot.bsky.social
AIエージェント同士は「会話」させない方がいい | sasakuna #zenn
https://zenn.dev/knowledgesense/articles/0464979e228c5c
AIエージェント同士は「会話」させない方がいい
zenn.dev
October 20, 2026 at 5:24 AM
technews4869.bsky.social
zenn.dev/knowledgesense/ar...
必要なToolだけをAgentに渡すことで46.9%のトークン消費を抑える方法
必要なToolだけをAgentに渡すことで46.9%のトークン消費を抑える方法
zenn.dev
October 10, 2026 at 8:43 AM
tech-trending.bsky.social
必要なToolだけをAgentに渡すことで46.9%のトークン消費を抑える方法
https://zenn.dev/knowledgesense/articles/f053192a51613f
必要なToolだけをAgentに渡すことで46.9%のトークン消費を抑える方法
zenn.dev
October 8, 2026 at 4:10 AM
zenn-ai-feed-bot.bsky.social
必要なToolだけをAgentに渡すことで46.9%のトークン消費を抑える方法 | sasakuna #zenn
https://zenn.dev/knowledgesense/articles/f053192a51613f
必要なToolだけをAgentに渡すことで46.9%のトークン消費を抑える方法
zenn.dev
October 6, 2026 at 5:25 PM
zenn-ai-feed-bot.bsky.social
RAG精度を自己進化させる方法。知能が安い時代に。 | Atsushi Kadowaki #zenn
https://zenn.dev/knowledgesense/articles/aa71237da8bd75
RAG精度を自己進化させる方法。知能が安い時代に。
zenn.dev
September 29, 2026 at 7:33 AM
zenn-ai-feed-bot.bsky.social
AI開発によって25倍に膨れたCI実行へのAnthropicの対処方法 | sasakuna #zenn
https://zenn.dev/knowledgesense/articles/445339edc2d499
AI開発によって25倍に膨れたCI実行へのAnthropicの対処方法
zenn.dev
September 22, 2026 at 5:27 AM
technews4869.bsky.social
zenn.dev/knowledgesense/ar...
SKILL.md を良くする技術。AIのブレをなくす。
SKILL.md を良くする技術。AIのブレをなくす。
zenn.dev
September 19, 2026 at 5:33 PM
viiiviii538.bsky.social
Skillの呼び出し精度がブレるの、結局書き方の問題だよなという記事。複数Skill運用してると「なぜこっちが呼ばれた」が増えるけど、原因の大半はSKILL.md側の記述の曖昧さにある。プロンプト力じゃなくて仕様書を書く力が要る、という話として読める。
https://zenn.dev/knowledgesense/articles/9adf1e9b17ffd6
September 18, 2026 at 3:38 AM
viiiviii538.bsky.social
SKILL.mdの記述次第で呼び出し精度とブレが変わるという実践記事。Skillを2つ3つ運用してると、微妙な説明文の差で意図しない方が呼ばれること結構あるから、今日読んで今日自分のSKILL.mdを直せる系の実用ネタだと思う。
https://zenn.dev/knowledgesense/articles/9adf1e9b17ffd6
September 17, 2026 at 3:47 AM
tech-trending.bsky.social
SKILL.md を良くする技術。AIのブレをなくす。
https://zenn.dev/knowledgesense/articles/9adf1e9b17ffd6
SKILL.md を良くする技術。AIのブレをなくす。
zenn.dev
September 16, 2026 at 6:50 AM
technews4869.bsky.social
zenn.dev/knowledgesense/ar...
Google提唱の「SKILL.state」について。プロンプトに型の概念を導入
Google提唱の「SKILL.state」について。プロンプトに型の概念を導入
zenn.dev
September 8, 2026 at 1:53 PM
viiiviii538.bsky.social
やっぱりそこに行き着いたか、という話。長時間動くエージェントが会話を重ねるほど劣化する問題に、スキルの状態を型として明示的に持たせるSKILL.stateという提案。今のSkillは基本ただの手順書なので、どこまで進んだかがモデルの記憶頼み。型で持てば途中再開と分岐管理が壊れにくくなる。
https://zenn.dev/knowledgesense/articles/ad123283bdea26
September 2, 2026 at 2:03 AM
tech-trending.bsky.social
Google提唱の「SKILL.state」について。プロンプトに型の概念を導入
https://zenn.dev/knowledgesense/articles/ad123283bdea26
Google提唱の「SKILL.state」について。プロンプトに型の概念を導入
zenn.dev
September 1, 2026 at 1:08 PM
newsen.bsky.social
Launch of Free Trial Campaign for Municipal 'Cowork' Service by KnowledgeSense#Japan#Tokyo#ChatSense#KnowledgeSense#Cowork
Launch of Free Trial Campaign for Municipal 'Cowork' Service by KnowledgeSense
KnowledgeSense has initiated a free trial campaign for its 'Cowork' service, aimed at municipalities, enabling secure AI integration into local government operations.
third-news.com
August 31, 2026 at 12:22 PM
newsen.bsky.social
Knowledge Sense's AI Service Enhances Corporate Data Management with New Notebook Feature#Japan#Tokyo#AI_service#ChatSense#KnowledgeSense
Knowledge Sense's AI Service Enhances Corporate Data Management with New Notebook Feature
Knowledge Sense is set to launch an innovative feature in its Notebook AI service, automatically generating data tables from corporate documents, enhancing efficiency.
third-news.com
August 26, 2026 at 3:35 AM
technews4869.bsky.social
zenn.dev/knowledgesense/ar...
敵対的検証をLLMコードレビューで試すと何が起きるのか
敵対的検証をLLMコードレビューで試すと何が起きるのか
zenn.dev
August 25, 2026 at 2:10 PM
tech-trending.bsky.social
敵対的検証をLLMコードレビューで試すと何が起きるのか
https://zenn.dev/knowledgesense/articles/0cb441713c9fee
敵対的検証をLLMコードレビューで試すと何が起きるのか
zenn.dev
August 25, 2026 at 1:20 PM
newsen.bsky.social
ChatSense Enhances Enterprise AI with Web Search Capabilities for Gemini and Claude Models#Japan#Tokyo#AI_Technologies#ChatSense#KnowledgeSense
ChatSense Enhances Enterprise AI with Web Search Capabilities for Gemini and Claude Models
KnowledgeSense has advanced its ChatSense enterprise AI platform to enable web searches for Gemini and Claude, allowing for real-time information retrieval.
third-news.com
August 25, 2026 at 9:22 AM
newsen.bsky.social
ChatSense Introduces AI Learning for Files Up to 1GB for Corporates#Japan#Tokyo#AI_Learning#ChatSense#KnowledgeSense
ChatSense Introduces AI Learning for Files Up to 1GB for Corporates
KnowledgeSense has enhanced its ChatSense platform, allowing companies to upload files as large as 1GB for AI learning, enhancing efficiency.
third-news.com
August 24, 2026 at 9:34 AM
newsen.bsky.social
New Cost Management Service for Corporate AI by ChatSense Aims to Streamline Expenses#Japan#Tokyo#AI_Management#ChatSense#KnowledgeSense
New Cost Management Service for Corporate AI by ChatSense Aims to Streamline Expenses
KnowledgeSense is set to launch a new service for ChatSense that allows corporations to manage AI costs effectively. This innovative solution promises efficiency improvements.
third-news.com
August 21, 2026 at 4:07 AM
newsen.bsky.social
KnowledgeSense Introduces Code Interpreter for ChatSense Using Gemini and Claude Models#Japan#Tokyo#Gemini#ChatSense#KnowledgeSense
KnowledgeSense Introduces Code Interpreter for ChatSense Using Gemini and Claude Models
KnowledgeSense enhances its AI agent, ChatSense, enabling code interpreter capabilities within Gemini and Claude models, promoting efficiency.
third-news.com
August 19, 2026 at 6:01 AM