<?xml version="1.0" encoding="UTF-8"?><rss version="2.0"><channel><title>saka2jp&apos;s blog</title><description>学びを深め、記憶に刻む — 技術・企業・書籍のDeepDiveブログ</description><link>https://saka2jp.vercel.app/</link><item><title>Paradox — 妻の名を冠した会話型AI「Olivia」でマクドナルド全店の採用を自動化、9年で$1B買収されたフロントライン採用の覇者、その光と影</title><link>https://saka2jp.vercel.app/deep-dive-company/paradox/</link><guid isPermaLink="true">https://saka2jp.vercel.app/deep-dive-company/paradox/</guid><description>2016年、元Jobing.com創業者Aaron Matosが妻の名「Olivia」を冠した会話型AIで創業。McDonald&apos;s・McHireをはじめFortune 500の時給労働採用を会話UXで寡占し、2021年に$1.5Bユニコーン化。2025年7月、McHireのパスワード&quot;123456&quot;で最大6,400万件の候補者データ露出という事件を経て、同年10月1日にWorkdayが$1B現金で買収完了。独立ユニコーンから巨艦Workdayの「Paradox @ Workday」へ——9年の軌跡とHR Tech再編の物語を徹底分析。</description><pubDate>Wed, 22 Apr 2026 00:00:00 GMT</pubDate><category>Paradox</category><category>Olivia</category><category>Conversational AI</category><category>McHire</category><category>Workday</category><category>Aaron Matos</category><category>Adam Godson</category><category>AI採用</category><category>HR Tech</category><category>フロントライン採用</category><category>High-Volume Hiring</category><category>IDOR</category><category>セキュリティ事故</category></item><item><title>React × TypeScriptで学ぶ無限スクロールの作法 — IntersectionObserverからTanStack Queryまで</title><link>https://saka2jp.vercel.app/tech-blog/react-typescript-infinite-scroll/</link><guid isPermaLink="true">https://saka2jp.vercel.app/tech-blog/react-typescript-infinite-scroll/</guid><description>React + TypeScriptで無限スクロールを実装する際の考え方・実装パターン・落とし穴を整理。IntersectionObserver基礎からカスタムフック化、useInfiniteQueryとの組み合わせまで実用的に解説します。</description><pubDate>Tue, 21 Apr 2026 00:00:00 GMT</pubDate><category>React</category><category>TypeScript</category><category>IntersectionObserver</category><category>TanStack Query</category><category>UX</category></item><item><title>Langfuse Deep Dive</title><link>https://saka2jp.vercel.app/deep-dive-tech/langfuse/</link><guid isPermaLink="true">https://saka2jp.vercel.app/deep-dive-tech/langfuse/</guid><description>YC W23発、2026年1月にClickHouseへ$15B評価で合流したOSS LLMエンジニアリングプラットフォーム「Langfuse」を10章で徹底解剖。Trace/Observation/Scoreのデータモデル、Postgres+ClickHouse+Redis+S3のv3アーキテクチャ、OpenTelemetryネイティブSDK、Prompt Management、LLM-as-a-Judge評価、Merck・LayerX・ZOZOの採用事例まで、2026年時点の最新実装知見を網羅</description><pubDate>Tue, 21 Apr 2026 00:00:00 GMT</pubDate><category>Langfuse</category><category>LLMOps</category><category>Observability</category><category>OpenTelemetry</category><category>LLM</category><category>AI</category><category>Prompt Management</category><category>Evaluation</category></item><item><title>HireVue — 累計8,000万人を面接したAI面接の発明者、顔分析廃止とMercor台頭の中で打つエージェント時代の一手</title><link>https://saka2jp.vercel.app/deep-dive-company/hirevue/</link><guid isPermaLink="true">https://saka2jp.vercel.app/deep-dive-company/hirevue/</guid><description>2004年にユタ州の大学生Mark Newmanが20歳で創業したHireVue。累計8,000万回の面接・Fortune 100の60%が採用する世界最大のAI面接プラットフォームは、2019年Carlyle Group傘下入り、2021年顔分析廃止、2023年Modern Hire買収、2026年Hireguide買収と激動の20年を経てきた。Mercor（評価額$10B）の台頭、EU AI Act 2026年8月の高リスクAI規制、ACLU差別告発——エージェント時代の生存戦略を徹底分析。</description><pubDate>Mon, 20 Apr 2026 00:00:00 GMT</pubDate><category>HireVue</category><category>AI面接</category><category>Carlyle Group</category><category>Modern Hire</category><category>Hireguide</category><category>Agentic AI</category><category>EU AI Act</category><category>EPIC</category><category>ORCAA</category><category>顔分析</category><category>TalentA</category><category>AI採用</category></item><item><title>『「本当にほしい人材」が集まる中途採用の定石』を読む — 採用を&quot;人気作り&quot;として再設計する</title><link>https://saka2jp.vercel.app/book-review/hontouni-hoshii-jinzai/</link><guid isPermaLink="true">https://saka2jp.vercel.app/book-review/hontouni-hoshii-jinzai/</guid><description>今啓亮『「本当にほしい人材」が集まる中途採用の定石 ― 採用広報からスカウト文章、面接術まで』を読んだ感想文。採用代行「まるごと人事」を運営する著者が提示する、売り手市場時代の採用戦略を、採用漏斗・KPI逆算思考・求人手法ポートフォリオ・スカウト文章・面接術の5レイヤーで整理する。エンジニア採用・転職者視点での示唆も併記。</description><pubDate>Sun, 19 Apr 2026 00:00:00 GMT</pubDate><category>中途採用</category><category>採用広報</category><category>スカウト</category><category>面接</category><category>人事</category><category>書評</category><category>HR</category><category>リクルーティング</category><category>マルゴト</category><category>今啓亮</category></item><item><title>SQL Deep Dive</title><link>https://saka2jp.vercel.app/deep-dive-tech/sql/</link><guid isPermaLink="true">https://saka2jp.vercel.app/deep-dive-tech/sql/</guid><description>1970年のCodd論文から2023年のSQL/PGQまで、半世紀にわたり事実上の標準であり続ける問い合わせ言語SQLを、設計哲学・関係代数・実行モデル・インデックス・トランザクション・方言・未来の10章で徹底解剖するシリーズ</description><pubDate>Sat, 18 Apr 2026 00:00:00 GMT</pubDate><category>SQL</category><category>データベース</category><category>RDBMS</category><category>PostgreSQL</category><category>MySQL</category><category>関係代数</category></item><item><title>Claude Opus 4.7リリース — SWE-bench首位奪還、xhighエフォート、API破壊的変更まとめ</title><link>https://saka2jp.vercel.app/tech-blog/claude-opus-4-7/</link><guid isPermaLink="true">https://saka2jp.vercel.app/tech-blog/claude-opus-4-7/</guid><description>2026年4月16日にリリースされたClaude Opus 4.7の全貌。SWE-bench Proで64.3%を記録しGPT-5.4・Gemini 3.1 Proを抑えてGA最強に。新エフォートレベル「xhigh」、タスクバジェット、3倍の画像解像度、/ultrareviewコマンド、そしてextended thinkingの廃止を含むAPI破壊的変更——開発者が知るべき全てをまとめる。</description><pubDate>Fri, 17 Apr 2026 00:00:00 GMT</pubDate><category>Claude</category><category>Anthropic</category><category>Opus 4.7</category><category>LLM</category><category>AI</category><category>ベンチマーク</category><category>SWE-bench</category><category>Claude Code</category><category>API</category></item><item><title>/ultrareview 完全ガイド — Claude Codeのマルチエージェント型バグハンティングを使いこなす</title><link>https://saka2jp.vercel.app/tech-blog/claude-ultrareview/</link><guid isPermaLink="true">https://saka2jp.vercel.app/tech-blog/claude-ultrareview/</guid><description>Claude Opus 4.7と同時にリリースされた/ultrareviewは、クラウド上で複数のレビュアーエージェントを並列展開し、発見したバグを独立検証してからレポートする「嘘をつかないコードレビュー」だ。従来の/reviewとの違い、マルチエージェント検証の仕組み、料金体系、そして3回の無料枠をどう使うべきかを解説する。</description><pubDate>Fri, 17 Apr 2026 00:00:00 GMT</pubDate><category>Claude Code</category><category>ultrareview</category><category>コードレビュー</category><category>Opus 4.7</category><category>Anthropic</category><category>AI</category><category>バグ検出</category><category>マルチエージェント</category></item><item><title>Mercor — 22歳のビリオネア3人が作った「AIが面接するプラットフォーム」、創業2年で評価額100億ドルの光と影</title><link>https://saka2jp.vercel.app/deep-dive-company/mercor/</link><guid isPermaLink="true">https://saka2jp.vercel.app/deep-dive-company/mercor/</guid><description>2023年、高校ディベート部の幼馴染3人が大学寮で創業したMercor。AIが20分のビデオ面接を行い、専門家をOpenAI・Anthropic等のAIラボにマッチングするプラットフォームは、創業17ヶ月でARR $500Mに到達し、22歳で世界最年少セルフメイドビリオネアを生んだ。しかし2026年3月、LiteLLMサプライチェーン攻撃で4TBのデータが流出。5件の集団訴訟、Fortune 500企業の契約解除——急成長の代償と、AI労働市場の未来を読み解く。</description><pubDate>Fri, 17 Apr 2026 00:00:00 GMT</pubDate><category>Mercor</category><category>AI面接</category><category>AI採用</category><category>RLHF</category><category>AIトレーニングデータ</category><category>Scale AI</category><category>Thiel Fellowship</category><category>サプライチェーン攻撃</category><category>データ侵害</category><category>ユニコーン</category></item><item><title>『図解 人材マネジメント入門』を読む — エンジニアが知っておくべきHRMの「6要素」と「2つの公平感」</title><link>https://saka2jp.vercel.app/book-review/zukai-jinzai-management/</link><guid isPermaLink="true">https://saka2jp.vercel.app/book-review/zukai-jinzai-management/</guid><description>坪谷邦生『図解 人材マネジメント入門 ― 人事の基礎をゼロからおさえておきたい人のための「理論と実践」100のツボ』を読んだ感想文。人材マネジメントを6要素で構造化し、人事評価を中心ハブに据える視点、「管理」と「マネジメント」の違い、分配と手続きの2つの公平感、4象限で企業を分類するフレームを、エンジニア視点で整理する。</description><pubDate>Fri, 17 Apr 2026 00:00:00 GMT</pubDate><category>人材マネジメント</category><category>HRM</category><category>人事</category><category>坪谷邦生</category><category>組織論</category><category>書評</category><category>エンジニアリングマネジメント</category></item><item><title>RAG / ベクトル検索 Deep Dive</title><link>https://saka2jp.vercel.app/deep-dive-tech/rag/</link><guid isPermaLink="true">https://saka2jp.vercel.app/deep-dive-tech/rag/</guid><description>LLMの限界を超える「検索拡張生成」を10章で徹底解剖。Embeddingsの原理、ANNアルゴリズム、pgvector実践、Advanced RAG（GraphRAG・Contextual Retrieval）、評価・運用まで2026年時点の最新知見を網羅</description><pubDate>Thu, 16 Apr 2026 00:00:00 GMT</pubDate><category>RAG</category><category>ベクトル検索</category><category>Embeddings</category><category>pgvector</category><category>LLM</category><category>AI</category></item><item><title>Chrome DevTools MCP徹底入門 — AIコーディングエージェントに「目」と「実ブラウザ」を与える</title><link>https://saka2jp.vercel.app/tech-blog/chrome-devtools-mcp/</link><guid isPermaLink="true">https://saka2jp.vercel.app/tech-blog/chrome-devtools-mcp/</guid><description>Chrome DevTools MCPは、Claude CodeなどのAIエージェントに実Chromeブラウザの操作・トレース記録・コンソール/ネットワーク検査能力を与えるMCPサーバー。29個のツール群の役割、セットアップ、パフォーマンス分析・デバッグ・Core Web Vitals監査の実践ワークフローを、公式情報ベースでまとめます。</description><pubDate>Wed, 15 Apr 2026 00:00:00 GMT</pubDate><category>MCP</category><category>Chrome DevTools</category><category>Claude Code</category><category>Performance</category><category>Debugging</category></item><item><title>Sakana AI — 進化的アプローチで挑む日本発Sovereign AIユニコーン、評価額4,320億円とTransformer共著者の&quot;第3の道&quot;</title><link>https://saka2jp.vercel.app/deep-dive-company/sakana-ai/</link><guid isPermaLink="true">https://saka2jp.vercel.app/deep-dive-company/sakana-ai/</guid><description>2023年7月、元Google Brainの David Ha、Transformer論文共著者のLlion Jones、元外交官の伊藤錬が創業した東京のAIラボ。「自然に学ぶ」進化的モデルマージを武器に、創業14ヶ月で日本最速ユニコーンを達成。2025年11月のSeries Bで$200M（約320億円）を調達し、post-money評価額4,320億円（$2.7B）は日本の非上場スタートアップ歴代最高。MUFG・ATLA・In-Q-Telを顧客/株主に据え、金融・防衛・政府の「Sovereign AI」市場で米中巨人と戦わない&quot;第3の道&quot;を切り拓く——Sakana AIの全貌に迫る。</description><pubDate>Wed, 15 Apr 2026 00:00:00 GMT</pubDate><category>Sakana AI</category><category>生成AI</category><category>LLM</category><category>基盤モデル</category><category>Sovereign AI</category><category>進化的アルゴリズム</category><category>AI Scientist</category><category>ユニコーン</category><category>David Ha</category><category>Transformer</category></item><item><title>Preferred Networks — 日本最大のAIユニコーンが挑む「AI半導体×基盤モデル×応用」の垂直統合戦略、MN-Core L1000で対NVIDIA推論10倍・PLaMoで日本語LLM世界最高クラスへ</title><link>https://saka2jp.vercel.app/deep-dive-company/preferred-networks/</link><guid isPermaLink="true">https://saka2jp.vercel.app/deep-dive-company/preferred-networks/</guid><description>2014年創業、東大ICPC出身の西川徹・岡野原大輔が率いる日本最大のAIユニコーン。AI半導体MN-Core→スーパーコンピュータMN-3（Green500首位3回）→基盤モデルPLaMo→産業応用までを垂直統合する、日本で唯一の「フルスタックAI企業」。累計調達300億円超、トヨタ・ファナック・ENEOS・SBI・三菱商事・講談社を株主に、GENIAC第1〜3期に連続採択。2025年11月に岡野原氏が社長CEOに昇格、MN-Core L1000（3D積層DRAMでGPU比推論10倍）とRapidus連携で「ソブリンAIインフラ」の主役を目指す——Preferred Networksの全貌に迫る。</description><pubDate>Tue, 14 Apr 2026 00:00:00 GMT</pubDate><category>Preferred Networks</category><category>PFN</category><category>AI半導体</category><category>MN-Core</category><category>PLaMo</category><category>生成AI</category><category>LLM</category><category>Matlantis</category><category>ユニコーン</category><category>GENIAC</category></item><item><title>PKSHA Technology — 東大松尾研発のアルゴリズム企業が、チャットボット・ボイスボット・FAQで国内シェア1位独占、AI SaaS ARR84億円とM&amp;Aで売上350億円への急成長シフト</title><link>https://saka2jp.vercel.app/deep-dive-company/pksha-technology/</link><guid isPermaLink="true">https://saka2jp.vercel.app/deep-dive-company/pksha-technology/</guid><description>2012年創業、東大松尾研発のAIアルゴリズム企業。AI Research&amp;Solution（59%）とAI SaaS（41%）の二本柱で、富士キメラ総研調査のチャットボット・ボイスボット・FAQの3部門で国内シェア1位を独占。FY2025売上217億円、AI SaaS ARR84億円・NRR104%を実現し、エクストーン・サーキュレーション・X Capital・Biz Freakの連続M&amp;Aでバリューチェーンを拡張。FY2026は売上350億円（+60.8%）の急成長フェーズへ——「未来のソフトウェアを形にする」アルゴリズム企業の全貌に迫る。</description><pubDate>Mon, 13 Apr 2026 00:00:00 GMT</pubDate><category>PKSHA Technology</category><category>AIエージェント</category><category>AI SaaS</category><category>自然言語処理</category><category>松尾研</category><category>チャットボット</category><category>VoiceAgent</category><category>LLM</category><category>RetNet</category><category>M&amp;A</category></item><item><title>CVE-2025-55182（React2Shell）徹底解析 — RSCデシリアライズRCEの仕組みと実践的防御設計</title><link>https://saka2jp.vercel.app/tech-blog/cve-2025-55182-rsc-rce/</link><guid isPermaLink="true">https://saka2jp.vercel.app/tech-blog/cve-2025-55182-rsc-rce/</guid><description>2025年12月に公開されたCVE-2025-55182（React2Shell）は、React Server ComponentsのFlightペイロードのデシリアライズ処理を突いた未認証RCE。CVSS 10.0の最高深刻度であり、Server Functionを1つも実装していないアプリでも影響を受けます。攻撃メカニズム・攻撃面の棚卸し・多層防御の実装・検知パターン・緊急アップデートのSLA設計まで、一次情報ベースで実践的にまとめます。</description><pubDate>Sun, 12 Apr 2026 00:00:00 GMT</pubDate><category>Security</category><category>React</category><category>RSC</category><category>CVE</category><category>Next.js</category><category>Incident Response</category></item><item><title>Claude Mythos Preview — 「一般提供されない」フロンティアモデルを読み解く</title><link>https://saka2jp.vercel.app/tech-blog/claude-mythos-preview/</link><guid isPermaLink="true">https://saka2jp.vercel.app/tech-blog/claude-mythos-preview/</guid><description>2026年4月7日にAnthropicが発表した新フロンティアモデル「Claude Mythos Preview」。Opus 4.6を大きく超える自律的な脆弱性発見・悪用能力を持ちながら、一般提供はされず「Project Glasswing」経由でのみ配布される異例のモデルです。一次情報から、何が違うのか・なぜ閉じたのか・一般開発者にとって何を意味するのかを読み解きます。</description><pubDate>Sat, 11 Apr 2026 00:00:00 GMT</pubDate><category>Claude</category><category>Anthropic</category><category>Frontier Model</category><category>Cybersecurity</category><category>Project Glasswing</category></item><item><title>Claude Managed Agents入門 — Anthropic純正の「マネージド・ハーネス」で長時間Agentを動かす</title><link>https://saka2jp.vercel.app/tech-blog/claude-managed-agents/</link><guid isPermaLink="true">https://saka2jp.vercel.app/tech-blog/claude-managed-agents/</guid><description>2026年4月にGAしたClaude Managed Agents。Messages APIとの違い、/v1/agents・/v1/sessionsの使い方、組み込みツールセット、そして「自前ハーネスを捨てる」判断基準まで実装例付きで解説します。</description><pubDate>Fri, 10 Apr 2026 00:00:00 GMT</pubDate><category>Claude</category><category>Anthropic API</category><category>AI Agent</category><category>Managed Agents</category></item><item><title>ハーネスエンジニアリング入門 — LLMを「賢く動かす」ための周辺装置を設計する</title><link>https://saka2jp.vercel.app/tech-blog/harness-engineering/</link><guid isPermaLink="true">https://saka2jp.vercel.app/tech-blog/harness-engineering/</guid><description>プロンプトでもモデル選定でもなく、LLMの「周辺装置」こそが性能を決める。ハーネスエンジニアリングの概念と、ツール設計・コンテキスト管理・フィードバックループの実践指針を解説します。</description><pubDate>Fri, 10 Apr 2026 00:00:00 GMT</pubDate><category>Harness Engineering</category><category>AI Agent</category><category>Claude Code</category><category>LLM</category></item><item><title>ラクスル — 印刷会社の&quot;非稼働時間&quot;を束ねたシェアリングPFが売上619億円・印刷EC首位に到達、ゴールドマン・サックスと1,300億円MBOで非公開化し「仕組みを変えれば、世界はもっと良くなる」の第二章へ</title><link>https://saka2jp.vercel.app/deep-dive-company/raksul/</link><guid isPermaLink="true">https://saka2jp.vercel.app/deep-dive-company/raksul/</guid><description>2009年創業、印刷会社の非稼働時間を活用するシェアリングPFで印刷EC首位に。ノバセル（広告）・ハコベル（物流）・ジョーシス（IT統制）と4事業を展開し売上619億円を達成。2026年3月にゴールドマン・サックスと約1,300億円のMBOが成立——「仕組みを変えれば、世界はもっと良くなる」を掲げるBtoBプラットフォーム企業の全貌に迫る。</description><pubDate>Fri, 10 Apr 2026 00:00:00 GMT</pubDate><category>ラクスル</category><category>Raksul</category><category>印刷EC</category><category>MBO</category><category>ノバセル</category><category>ハコベル</category><category>ジョーシス</category><category>シェアリングエコノミー</category><category>BtoBプラットフォーム</category></item><item><title>LayerX — Gunosy創業者がブロックチェーンからSaaSへ大胆ピボット、「バクラク」×「ALTERNA」×「Ai Workforce」の三刀流でARR100億円・評価額1,000億円に到達した&quot;コンパウンドスタートアップ&quot;の全貌</title><link>https://saka2jp.vercel.app/deep-dive-company/layerx/</link><guid isPermaLink="true">https://saka2jp.vercel.app/deep-dive-company/layerx/</guid><description>東大松尾研からGunosyを創業・上場させた福島良典が、2018年にブロックチェーン企業として設立したLayerX。「日本はブロックチェーンどころではない」と気づき法人支出管理SaaS「バクラク」へ大胆にピボット。三井物産との合弁でデジタル証券「ALTERNA」、さらにエンタープライズAI「Ai Workforce」を展開し、わずか5年でARR100億円・累計調達282億円・推定評価額1,000億円に到達。行動指針に「徳」と「Bet AI」を掲げ、2030年ARR1,000億円を目指す&quot;コンパウンドスタートアップ&quot;の全貌に迫る。</description><pubDate>Thu, 09 Apr 2026 00:00:00 GMT</pubDate><category>LayerX</category><category>バクラク</category><category>SaaS</category><category>AI</category><category>Fintech</category><category>コンパウンドスタートアップ</category><category>LLM</category><category>デジタル証券</category></item><item><title>Sansan — 名刺管理のパイオニアがシェア86%の営業DXから請求書・契約・AIデータ基盤へ拡張、連結ARR459億円・時価総額1,600億円に到達した&quot;ビジネスインフラ&quot;の全貌</title><link>https://saka2jp.vercel.app/deep-dive-company/sansan/</link><guid isPermaLink="true">https://saka2jp.vercel.app/deep-dive-company/sansan/</guid><description>2007年創業、名刺管理SaaS市場でシェア86%を誇るSansan。Bill Oneでインボイス管理シェア49%を獲得し、Contract One・Data Intelligenceへと拡張。DXからAXへ——「出会いからイノベーションを生み出す」ビジネスインフラ企業の全貌に迫る。</description><pubDate>Thu, 09 Apr 2026 00:00:00 GMT</pubDate><category>Sansan</category><category>Bill One</category><category>Contract One</category><category>名刺管理</category><category>SaaS</category><category>AI</category><category>営業DX</category><category>インボイス制度</category></item><item><title>カオナビ — 「顔が見える人事」で8年連続シェアNo.1を獲得したタレマネSaaSが、カーライル497億円TOBで非公開化し&quot;Talent intelligence&quot;への進化を賭ける全貌</title><link>https://saka2jp.vercel.app/deep-dive-company/kaonavi/</link><guid isPermaLink="true">https://saka2jp.vercel.app/deep-dive-company/kaonavi/</guid><description>アクセンチュア出身の柳橋仁機が2008年に創業し、「顔写真ベースの直感的UI」という独自のプロダクト思想で日本のタレントマネジメントシステム市場を切り拓いたカオナビ。利用企業数4,500社超・ARR100億円突破・月次解約率0.40%という卓越したSaaS KPIを達成し、8年連続市場シェアNo.1を維持する。2025年6月、米PEファンドのカーライル・グループが約497億円の「借金なし・のれんなし」TOBで非公開化。新ビジョン「Talent intelligence」のもと、マルチプロダクト戦略とAI投資を加速し、5年後の再上場で数千億円企業を目指す挑戦の全貌に迫る。</description><pubDate>Wed, 08 Apr 2026 00:00:00 GMT</pubDate><category>カオナビ</category><category>Kaonavi</category><category>HR Tech</category><category>タレントマネジメント</category><category>SaaS</category><category>AI</category><category>カーライル</category><category>TOB</category></item><item><title>タレントパレット — NRI出身の創業者がテキストマイニング技術で「科学的人事」を切り拓く、営業利益率37%超のHR Techプラットフォーム</title><link>https://saka2jp.vercel.app/deep-dive-company/talent-palette/</link><guid isPermaLink="true">https://saka2jp.vercel.app/deep-dive-company/talent-palette/</guid><description>野村総合研究所出身の三室克哉が2006年に創業したプラスアルファ・コンサルティング。テキストマイニングSaaS「見える化エンジン」で14年連続シェアNo.1を獲得した後、その分析技術をHR領域に転用して2016年にリリースした「タレントパレット」が爆発的に成長。契約社数2,000超・継続率99.6%・ARPU月額45.9万円という驚異的なKPIを誇り、売上高170億円・営業利益率37.3%を達成。マイナビ・ラクスとの資本業務提携で採用からタレマネまでのフルカバレッジ戦略を加速させる「見える化プラットフォーム企業」の全貌に迫る。</description><pubDate>Wed, 08 Apr 2026 00:00:00 GMT</pubDate><category>タレントパレット</category><category>プラスアルファ・コンサルティング</category><category>HR Tech</category><category>タレントマネジメント</category><category>SaaS</category><category>AI</category><category>テキストマイニング</category></item><item><title>ドメイン駆動設計（DDD） Deep Dive</title><link>https://saka2jp.vercel.app/deep-dive-tech/ddd/</link><guid isPermaLink="true">https://saka2jp.vercel.app/deep-dive-tech/ddd/</guid><description>ドメイン駆動設計（DDD）の戦略的設計から戦術的設計、CQRS・イベントソーシング、TypeScript実装、AI時代の展望まで、10章で体系的に徹底解剖するシリーズ</description><pubDate>Wed, 08 Apr 2026 00:00:00 GMT</pubDate><category>DDD</category><category>ドメイン駆動設計</category><category>設計</category><category>アーキテクチャ</category><category>ソフトウェア設計</category></item><item><title>Claude Code設定のベストプラクティス — CLAUDE.md・ルール・レビューエージェント完全ガイド</title><link>https://saka2jp.vercel.app/tech-blog/claude-code-configuration/</link><guid isPermaLink="true">https://saka2jp.vercel.app/tech-blog/claude-code-configuration/</guid><description>CLAUDE.mdの書き方、.claude/rules/によるモジュール化、Hooks・Skills・レビューエージェントの設定まで、Claude Codeの設定ベストプラクティスを実践例つきで解説します。</description><pubDate>Mon, 06 Apr 2026 00:00:00 GMT</pubDate><category>Claude Code</category><category>AI</category><category>開発環境</category><category>ベストプラクティス</category></item><item><title>Sentry完全活用ガイド2026 — アラート設計・AI機能Seer・最新トピック</title><link>https://saka2jp.vercel.app/tech-blog/sentry-2026/</link><guid isPermaLink="true">https://saka2jp.vercel.app/tech-blog/sentry-2026/</guid><description>Sentryのアラートルール最適化、AIデバッグエージェント「Seer」のMCP対応・ローカル連携、Log Drains、OpenTelemetry統合など2026年最新アップデートを包括的に解説します。</description><pubDate>Mon, 06 Apr 2026 00:00:00 GMT</pubDate><category>Sentry</category><category>モニタリング</category><category>エラー監視</category><category>AI</category><category>DevOps</category></item><item><title>Voice AIエージェントフレームワーク徹底比較 — LiveKit / Pipecat / Vocode / VAPI</title><link>https://saka2jp.vercel.app/tech-blog/voice-ai-framework-comparison/</link><guid isPermaLink="true">https://saka2jp.vercel.app/tech-blog/voice-ai-framework-comparison/</guid><description>Voice AIエージェント構築の主要フレームワーク4つを設計思想・DX・拡張性・ユースケースの観点で徹底比較します。</description><pubDate>Sun, 05 Apr 2026 00:00:00 GMT</pubDate><category>Voice AI</category><category>LiveKit</category><category>Pipecat</category><category>フレームワーク比較</category></item><item><title>Voice AIエージェントをゼロから作る — WebSocket + STT + LLM + TTS 構築記</title><link>https://saka2jp.vercel.app/tech-blog/voice-ai-from-scratch/</link><guid isPermaLink="true">https://saka2jp.vercel.app/tech-blog/voice-ai-from-scratch/</guid><description>フレームワークなしでVoice AIエージェントをゼロから構築。音声パイプラインの各ステージを実装しながら、その複雑さと勘所を学びます。</description><pubDate>Sat, 04 Apr 2026 00:00:00 GMT</pubDate><category>Voice AI</category><category>WebSocket</category><category>ハンズオン</category><category>STT</category><category>TTS</category></item><item><title>ターンテイキングの科学 — Voice AIが人間らしく会話するための技術</title><link>https://saka2jp.vercel.app/tech-blog/voice-ai-turn-taking/</link><guid isPermaLink="true">https://saka2jp.vercel.app/tech-blog/voice-ai-turn-taking/</guid><description>人間の話者交代は平均200msの精度。Voice AIが自然な会話を実現するためのターンテイキング技術を科学的知見から解説します。</description><pubDate>Fri, 03 Apr 2026 00:00:00 GMT</pubDate><category>Voice AI</category><category>ターンテイキング</category><category>VAD</category><category>会話AI</category><category>自然言語処理</category></item><item><title>パイプライン vs エンドツーエンド — Voice AIアーキテクチャの二大潮流</title><link>https://saka2jp.vercel.app/tech-blog/voice-ai-pipeline-vs-e2e/</link><guid isPermaLink="true">https://saka2jp.vercel.app/tech-blog/voice-ai-pipeline-vs-e2e/</guid><description>Voice AIエージェントの二大アーキテクチャ、パイプライン型とエンドツーエンド型を徹底比較。トレードオフと選定指針を解説します。</description><pubDate>Thu, 02 Apr 2026 00:00:00 GMT</pubDate><category>Voice AI</category><category>アーキテクチャ</category><category>GPT-4o</category><category>Gemini Live</category><category>Speech-to-Speech</category></item><item><title>200msの壁 — Voice AIエージェントのレイテンシ設計</title><link>https://saka2jp.vercel.app/tech-blog/voice-ai-latency-budget/</link><guid isPermaLink="true">https://saka2jp.vercel.app/tech-blog/voice-ai-latency-budget/</guid><description>人間の会話の応答速度は平均200ms。Voice AIエージェントがこの壁を超えるためのレイテンシバジェット設計と最適化手法を解説します。</description><pubDate>Wed, 01 Apr 2026 00:00:00 GMT</pubDate><category>Voice AI</category><category>レイテンシ</category><category>リアルタイム処理</category><category>WebRTC</category></item><item><title>HERP — 「採用を変え、日本を強く。」スクラム採用の提唱者が目指すコンパウンドHRプラットフォーム</title><link>https://saka2jp.vercel.app/deep-dive-company/herp/</link><guid isPermaLink="true">https://saka2jp.vercel.app/deep-dive-company/herp/</guid><description>リクルート→エウレカを経た庄田一郎が2017年に創業したHERP。「スクラム採用」という概念を提唱し、全社員参加型の採用管理システムHERP Hireで3,000社超の導入を達成。Haskellバックエンド・Kubernetes基盤という技術的先進性を武器に、AI Recruiter・HERP Career・ジョブミルを次々投入し、採用企業・人材紹介会社・求職者の三方向をつなぐコンパウンドHRプラットフォームへの進化を加速させる。</description><pubDate>Wed, 01 Apr 2026 00:00:00 GMT</pubDate><category>HERP</category><category>HR Tech</category><category>ATS</category><category>スクラム採用</category><category>AI</category><category>スタートアップ</category></item><item><title>ビズリーチ — 「新しい可能性を、次々と。」モルガン・スタンレー出身の南壮一郎が創った、日本のダイレクトリクルーティング革命</title><link>https://saka2jp.vercel.app/deep-dive-company/bizreach/</link><guid isPermaLink="true">https://saka2jp.vercel.app/deep-dive-company/bizreach/</guid><description>モルガン・スタンレー→楽天イーグルス創設メンバーという異色の経歴を持つ南壮一郎が2007年に創業したビズリーチ。日本初の「求職者課金型」転職プラットフォームで市場を切り開き、スカウト可能会員307万人超・導入企業3.8万社を擁するダイレクトリクルーティングのNo.1企業に成長。親会社Visionalは売上高801億円・営業利益214億円を達成し、生成AI特許数日本1位のBizReach AIとThinkings買収で「キャリアインフラ」への進化を加速させる。</description><pubDate>Tue, 31 Mar 2026 00:00:00 GMT</pubDate><category>ビズリーチ</category><category>Visional</category><category>HR Tech</category><category>ダイレクトリクルーティング</category><category>SaaS</category><category>AI</category></item><item><title>LLM Deep Dive</title><link>https://saka2jp.vercel.app/deep-dive-tech/llm/</link><guid isPermaLink="true">https://saka2jp.vercel.app/deep-dive-tech/llm/</guid><description>大規模言語モデル（LLM）の仕組みから最新動向まで、Transformerアーキテクチャ・訓練・推論・安全性・エコシステムを10章で徹底解剖するシリーズ</description><pubDate>Tue, 31 Mar 2026 00:00:00 GMT</pubDate><category>LLM</category><category>大規模言語モデル</category><category>Transformer</category><category>AI</category><category>深層学習</category></item><item><title>TypeScript 6.0の全貌 — 新機能・破壊的変更・移行ガイド</title><link>https://saka2jp.vercel.app/tech-blog/typescript-6/</link><guid isPermaLink="true">https://saka2jp.vercel.app/tech-blog/typescript-6/</guid><description>TypeScript 6.0の新機能、破壊的変更、5.xからの移行手順を包括的に解説。Go製コンパイラへの橋渡しとなるこのリリースで何が変わり、何を準備すべきかを整理します。</description><pubDate>Mon, 30 Mar 2026 00:00:00 GMT</pubDate><category>TypeScript</category><category>JavaScript</category><category>マイグレーション</category><category>コンパイラ</category></item><item><title>マネーフォワード — 「お金を前へ。人生をもっと前へ。」家計簿アプリから始まった、No.1バックオフィスAIカンパニーへの挑戦</title><link>https://saka2jp.vercel.app/deep-dive-company/moneyforward/</link><guid isPermaLink="true">https://saka2jp.vercel.app/deep-dive-company/moneyforward/</guid><description>ソニー・マネックス証券を経た辻庸介が創業したマネーフォワード。家計簿アプリ「ME」1,780万ユーザーを起点に法人向けSaaSへ展開し、売上503億円・SaaS ARR 393億円を達成。2025年11月期に創業初の最終黒字化を果たし、「No.1バックオフィスAIカンパニー」への変革を加速する軌跡と戦略を包括的に分析する。</description><pubDate>Mon, 30 Mar 2026 00:00:00 GMT</pubDate><category>マネーフォワード</category><category>SaaS</category><category>クラウド会計</category><category>フィンテック</category><category>AI</category><category>バックオフィス</category></item><item><title>Next.js Deep Dive</title><link>https://saka2jp.vercel.app/deep-dive-tech/nextjs/</link><guid isPermaLink="true">https://saka2jp.vercel.app/deep-dive-tech/nextjs/</guid><description>Pages RouterからApp Router、Server ComponentsからCache Componentsまで、Reactフルスタックフレームワークの全貌を10章で徹底解剖するシリーズ</description><pubDate>Sun, 29 Mar 2026 00:00:00 GMT</pubDate><category>Next.js</category><category>React</category><category>フロントエンド</category><category>フルスタック</category><category>Vercel</category></item><item><title>freee — 「スモールビジネスを、世界の主役に。」統合型クラウドERPで挑むAIネイティブカンパニーへの進化</title><link>https://saka2jp.vercel.app/deep-dive-company/freee/</link><guid isPermaLink="true">https://saka2jp.vercel.app/deep-dive-company/freee/</guid><description>Google出身の佐々木大輔が創業したfreee。クラウド会計ソフトから統合型ERPへ、そしてAIネイティブカンパニーへ。ARR344億円・有料課金60万社超の軌跡と、AIエージェント時代の戦略を包括的に分析する。</description><pubDate>Sat, 28 Mar 2026 00:00:00 GMT</pubDate><category>freee</category><category>SaaS</category><category>クラウド会計</category><category>ERP</category><category>AI</category><category>フィンテック</category></item><item><title>TypeScript Deep Dive</title><link>https://saka2jp.vercel.app/deep-dive-tech/typescript/</link><guid isPermaLink="true">https://saka2jp.vercel.app/deep-dive-tech/typescript/</guid><description>TypeScriptの設計哲学からコンパイラ内部構造、型システムの基礎と応用、2026年のエコシステム、Go移植によるProject Corsaまで、Web開発のデファクトスタンダードを10章で徹底解剖するシリーズ</description><pubDate>Fri, 27 Mar 2026 00:00:00 GMT</pubDate><category>TypeScript</category><category>JavaScript</category><category>型システム</category><category>コンパイラ</category><category>フロントエンド</category></item><item><title>Astro Islands Architectureの仕組みと活用法</title><link>https://saka2jp.vercel.app/tech-blog/astro-islands-architecture/</link><guid isPermaLink="true">https://saka2jp.vercel.app/tech-blog/astro-islands-architecture/</guid><description>Astroが採用するIslands Architectureの仕組みを解説し、パフォーマンス最適化の観点から活用法を紹介します。</description><pubDate>Thu, 26 Mar 2026 00:00:00 GMT</pubDate><category>Astro</category><category>Islands Architecture</category><category>パフォーマンス</category><category>React</category></item><item><title>SmartHR — 日本のHR Techユニコーンが描く「人的資本経営プラットフォーム」への進化</title><link>https://saka2jp.vercel.app/deep-dive-company/smarthr/</link><guid isPermaLink="true">https://saka2jp.vercel.app/deep-dive-company/smarthr/</guid><description>ARR200億円超・登録7万社のHR Techユニコーン SmartHR。労務管理クラウドから人的資本経営プラットフォームへの進化戦略を包括的に分析する。</description><pubDate>Thu, 26 Mar 2026 00:00:00 GMT</pubDate><category>SmartHR</category><category>HR Tech</category><category>SaaS</category><category>ユニコーン</category><category>人的資本経営</category></item><item><title>React Deep Dive</title><link>https://saka2jp.vercel.app/deep-dive-tech/react/</link><guid isPermaLink="true">https://saka2jp.vercel.app/deep-dive-tech/react/</guid><description>Reactの設計哲学からアーキテクチャ、最新のServer Components・Compilerまで、フロントエンド最大のUIライブラリを10章で徹底解剖するシリーズ</description><pubDate>Thu, 26 Mar 2026 00:00:00 GMT</pubDate><category>React</category><category>フロントエンド</category><category>JavaScript</category><category>TypeScript</category><category>UI</category></item></channel></rss>