[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"blogCategories:ja":3,"blog::market-insights:ja":28},[4,7,10,13,16,19,22,25],{"title":5,"slug":6},"Event Report","eventreport",{"title":8,"slug":9},"Product Guide","product-guide",{"title":11,"slug":12},"Tips & Case Study","tips-case-study",{"title":14,"slug":15},"Vision","vision",{"title":17,"slug":18},"エンジニアリング","engineering",{"title":20,"slug":21},"プロダクト","product",{"title":23,"slug":24},"プロダクト比較","product-comparison",{"title":26,"slug":27},"マーケットインサイト","market-insights",[29,42,52,62,72,82,92],{"title":30,"slug":31,"description":32,"author":33,"category":34,"coverImageUrl":36,"ogImageUrl":37,"createdAt":38,"updatedAt":39,"datePublished":40,"locale":41},"ソフトウェアを「どの深さまで」生成させるか — 適合と負担のトレードオフ","dynamic-generation-boundary","ソフトウェアを顧客ごとにどこまで深く生成するかは、適合と「抱える負担」のトレードオフ。UI・app・db・infra\u002Fauth の各層で、深く生成するほど自社への適合は増すが、構築・運用・セキュリティ・保守の負担も増す。エージェントが下げるのは構築コストで、運用・保守の負担は残る。Salesforce・Notion・Palantir の違いを「どの層を固定し、どの層を生成するか」で読み解き、データ分析でのアナロジーまで、どこまで生成すべきかの枠組みを示す。","Naoki Shibayama",{"title":26,"slug":27,"description":35},"","https:\u002F\u002Fimages.ctfassets.net\u002Fggtw2zqmifs5\u002FQHcCTFntE5dCFaWlkIRq6\u002F3f9e09bca03d181a8f680fc9b365d340\u002Fdynamic-generation-boundary-cover-blueprint.png","https:\u002F\u002Fimages.ctfassets.net\u002Fggtw2zqmifs5\u002F7ensVq3DvkDku6miK4tuAF\u002F10aa3e9e7017952cdd8ea42e23d05b0c\u002Fdynamic-generation-boundary-hero-blueprint-fixed.png","2026-06-09T01:00:10.816Z","2026-06-16T03:05:20.410Z","2026-06-11T08:30+09:00","ja",{"title":43,"slug":44,"description":45,"author":33,"category":46,"coverImageUrl":47,"ogImageUrl":48,"createdAt":49,"updatedAt":50,"datePublished":51,"locale":41},"「人間の」意思決定までもAIに任せられるようになるのか","agent-orchestrator-unit","AIに作業を任せても楽にならないのはなぜか。増えているのは「判断」だ、という観点から、判断は人間に残るとする見方（Nova Spivackの\"The Orchestrator\"）と、評価や実務をAIに渡す動き（LLM-as-a-judge・τ-bench・Constitutional AI・AlphaEvolve）を整理し、意思決定を「上（目的設定）」と「下（細かい判断）」に分けて考えます。",{"title":26,"slug":27,"description":35},"https:\u002F\u002Fimages.ctfassets.net\u002Fggtw2zqmifs5\u002F6Inwc4RDQBMptAMzjs96Gs\u002Fce704cfbfda741de21853c068d2f79d9\u002Fagent-orchestrator-unit-cover-blueprint.png","https:\u002F\u002Fimages.ctfassets.net\u002Fggtw2zqmifs5\u002F63KIUwzvi7m0c9NA4zdbIS\u002Fc9576f265084fd6c85da393159baffd6\u002Fagent-orchestrator-unit-hero-blueprint-fixed.png","2026-06-04T01:00:05.117Z","2026-06-16T03:05:26.133Z","2026-06-04T10:00+09:00",{"title":53,"slug":54,"description":55,"author":33,"category":56,"coverImageUrl":57,"ogImageUrl":58,"createdAt":59,"updatedAt":60,"datePublished":61,"locale":41},"再評価される Push 型 BI — 休みでも気づいてくれる LLM Agent","push-bi-revival","Tableau Pulse \u002F Power BI Copilot \u002F Looker (Gemini) の Push 型 BI 機能を、2016 年 Mike Smitheman の Push Intelligence 提唱まで遡って整理。コンセプトは 10 年前から完成済みで、動き出したのは周辺の前提（Slack 遍在化 \u002F LLM コスト下落 \u002F 自然言語要約 \u002F 双方向 followup）が揃ったからという構造を読み解きます。",{"title":26,"slug":27,"description":35},"https:\u002F\u002Fimages.ctfassets.net\u002Fggtw2zqmifs5\u002F4nR1TyJcxZ3sA6yCKa3agU\u002F0427372c0492669734b4f763d3848211\u002Fcover.png","https:\u002F\u002Fimages.ctfassets.net\u002Fggtw2zqmifs5\u002F37bHi4cyLug0uRa73kAHVB\u002Fd17d4a33a88ded0c8cbce37a775c906d\u002Fhero.png","2026-05-19T00:00:16.503Z","2026-05-21T09:46:15.188Z","2026-05-19T09:00:00+09:00",{"title":63,"slug":64,"description":65,"author":33,"category":66,"urlCategorySlug":21,"coverImageUrl":67,"ogImageUrl":68,"createdAt":69,"updatedAt":70,"datePublished":71,"locale":41},"内製データ Agent の現在地 — OpenAI と Meta が見せた \"6 層の context\"","in-house-data-agent","OpenAI \u002F Meta が公開した内製データ Agent の構造（OpenAI 6 層 \u002F Meta Cookbook-Recipe-Ingredient）を、dbt SoAE 2026 の業界数字と並べて整理。\"context engineering\" は schema 以外の組織知を retrievable に整理する地味な作業、というのが核。中堅企業向け Skeptics 視点も提示。",{"title":26,"slug":27,"description":35},"https:\u002F\u002Fimages.ctfassets.net\u002Fggtw2zqmifs5\u002F2cnf7pCQu7OQdiHrnea0NN\u002F345e3a5a0bf68af3097eecb51833c72e\u002Fcover-blueprint.png","https:\u002F\u002Fimages.ctfassets.net\u002Fggtw2zqmifs5\u002F4S5t1uJmGZRVH0JXx92LAy\u002Fe9ea86158fc39f04cef5e2c08effa752\u002Fhero-blueprint.png","2026-05-12T00:00:10.863Z","2026-05-21T09:46:17.058Z","2026-05-12T09:00:00+09:00",{"title":73,"slug":74,"description":75,"author":33,"category":76,"urlCategorySlug":21,"coverImageUrl":77,"ogImageUrl":78,"createdAt":79,"updatedAt":80,"datePublished":81,"locale":41},"Modern Data Stack の \"終焉\" と再構成 — dbt × Fivetran 合併が示すもの","modern-data-stack-2026","2025-10 dbt × Fivetran 合併と、前後 10 ヶ月で集中した 7 件の M&A を Tristan Handy \u002F 各社公式 \u002F Modern Data 101 の論で整理。best-of-breed の境界が溶け、agent \u002F context layer が次の差別化点に — という構造変化を curatorial に解説。中堅企業向け Skeptics 視点も提示。",{"title":26,"slug":27,"description":35},"https:\u002F\u002Fimages.ctfassets.net\u002Fggtw2zqmifs5\u002F2cjHUHQyXVEQDs1sW9jy59\u002F5b111b484f6cbaf213f7becb552b9394\u002Fcover-blueprint.png","https:\u002F\u002Fimages.ctfassets.net\u002Fggtw2zqmifs5\u002F2H0FOVXg1Mr5PIeZKYv0W0\u002F233cb95399cb03fe204019f455a35582\u002Fhero-blueprint.png","2026-05-08T00:00:09.700Z","2026-05-22T06:15:05.180Z","2026-05-08T09:00:00+09:00",{"title":83,"slug":84,"description":85,"author":33,"category":86,"urlCategorySlug":21,"coverImageUrl":87,"ogImageUrl":88,"createdAt":89,"updatedAt":90,"datePublished":91,"locale":41},"Open Semantic Interchange (OSI) を読み解く — Snowflake \u002F dbt \u002F Databricks が同じ規格に賛同した理由","open-semantic-interchange-2026","2026 年 1 月、Open Semantic Interchange (OSI) v1.0 が公開され、Snowflake \u002F dbt \u002F Databricks など 30+ ベンダーが同じセマンティックレイヤー規格に賛同しました。なぜこの標準化が今起きたのか、YAML 仕様の中身（ai_context フィールド含む）、各ベンダーの動機、競合する LookML \u002F Cube \u002F Snowflake Semantic View との関係を整理します。",{"title":26,"slug":27,"description":35},"https:\u002F\u002Fimages.ctfassets.net\u002Fggtw2zqmifs5\u002F4Kf5zOZf2BytPX1OC96EJ5\u002Fd5b5f20fc57852d6a319811cc980ba2e\u002Fcover.png","https:\u002F\u002Fimages.ctfassets.net\u002Fggtw2zqmifs5\u002F3CYTBddNeVh240GSpiwM40\u002F6469521dd52ee8ac385d36b2e0c9bf9e\u002Fhero.png","2026-04-30T23:00:12.364Z","2026-05-22T06:15:06.847Z","2026-05-01T08:00+09:00",{"title":93,"slug":94,"description":95,"author":33,"category":96,"urlCategorySlug":21,"coverImageUrl":97,"ogImageUrl":98,"createdAt":99,"updatedAt":100,"datePublished":101,"locale":41},"意味からコンテキストへ — Data Agent が本当に必要としているもの","context-layer-data-analytics","Data Agent の登場により、Semantic Layer だけではデータ分析の精度が出ないことが明らかになりつつあります。業界が注目する Context Layer とは何か、その背景と私たちの考えをまとめました。",{"title":26,"slug":27,"description":35},"https:\u002F\u002Fimages.ctfassets.net\u002Fggtw2zqmifs5\u002F1Ze3g4G0XQqkFGmlmQVyyU\u002F91cff72f0acf17d663b8d4f096ef145c\u002Fcover.jpg","https:\u002F\u002Fimages.ctfassets.net\u002Fggtw2zqmifs5\u002F6GMKVlHNNgJ8lqZyvb6Qj7\u002Fa1620910ba10f4fd04fe993707649bfc\u002Fog.jpg","2026-04-15T08:58:33.415Z","2026-05-07T15:35:38.456Z","2026-04-16T10:00+09:00"]