Tatsuma Murase

"I forgot I was even operating a PTZ." That's a quote from one of our operators during the DIG CUP beach volleyball broadcast on March 14-15, captured in a new case study from RCT (Remote Camera Technology). When a tool becomes invisible to the person using it, the work gets better. True of cameras. True of AI. Rarely talked about with the weight it deserves. A few things made this broadcast interesting to me: Sand court, no fixed seating, no permanent camera platforms, no shelter. Multi-cam AR production from temporary scaffolding and tents. AI-based real-time ball tracking, with AR objects correctly occluded by the ball. Ball trajectory lines for in/out calls during replays. The team operated PTZ cameras remotely using RCT's FR-2 controller. The framing comment in the case study, "Without the FR-2, we wouldn't have taken this job," says everything about how much the tool stack determines what's actually shootable. Same pattern I keep seeing across our AI rollout: better tooling doesn't just make existing work faster, it unlocks work that wasn't previously feasible. Sports, AR, AI, and remote control are converging fast, and the crews that can move with that pace get to define what live sports looks like next. Huge respect to the RCT team, and to our production crew for pulling this off in the sand.

“生成AIを活用したユーザー参加型オフライン体験コンテンツ” です！！ ボール絵になって入り込むぞー! -- モンスト、AIで“自分がモンスターになる”体験を提供 mixi.co.jp/news/2026/0514…

Been waiting for this. @runwayml is coming to Japan. At MIXI, Runway has already been part of our production workflow for a while. Not a pilot. A daily tool. The exciting part isn't the announcement. It's that Runway keeps evolving fast, and our workflow gets to keep evolving with it. runwayml.com/news/runway-is…

#Vaultroom シール！何に貼ろうかなー

youtu.be/wojvsOIuyKU?si… FC東京ファンとか、千葉ジェッツファン、スポーツファンの方はぜひコレを一回見てみてほしい！！ MIXIがTIPSTARなどで培ってきた、 AIを使ったLIVE映像演出の技術 色んなスポーツのリプレイなどもメッチャわかりやすく作れると思う。 TV局さん、DAZNさん、うちの技術導入しません？ （╹◡╹）

この間にビーチバレーの大会、「BV.LEAGUE DIG CUP Presented by MIXI」で使った映像配信のハイライトです。ARでの広告や、3D 合成や軌道など、視聴体験を向上させるためのアウトプットをこれからも沢山行っていきます！ ＜HIGH LIGHT Short ver.＞ 2026 BV.LEAGUE DIG CUP Presented by MIXI youtu.be/wojvsOIuyKU?si… via @YouTube

非常に自由度の高い福利厚生にしましたー！AIのサブスクも使えるー！ -- MIXI、独自枠「発明補助」を含む年24万円分の新福利厚生を開始 prtimes.jp/main/html/rd/p… via @PRTIMES_JP

One month after AI DAY 2026, the question changed inside MIXI. Time to Market measurement is now mandatory on every project. The quality bar stays the same, or rises. AI makes more iterations and more reviews cheap, so we expect both axes to move. Last year's question was "Are you using AI?" 99% adoption made that question useless. This year's question is "How much faster did you actually ship, without dropping the bar?" Every team brings TTM numbers to operating reviews. Idea to production. Spec to release. The unit is up to the team. Measurement is not. What we're learning: high AI adoption doesn't automatically shorten TTM. Some teams use AI a lot, but the workflow around it never changed. 99% adoption was a milestone, not the destination. The destination is shipping faster, at the same quality bar or higher. Speed alone is just faster output, not progress.

岐阜県の麗澤瑞浪高等学校さんに登壇行ったときのレポートが上がっていました。100人の大人と会うという素敵なプロジェクトの1人目として、これから99人の大人たちが独自のこと "だけ" を喋れるように、みんながいいそうなことはある程度全部言っておきました！

何故にこういうことをしているかは、ここでちょうど会話してきました。 youtu.be/XieRv0QoRDU

渋谷にRePlayce さんのHR高等学院と共に「SHIBUYA MIXI BASE」の誕生でございます。AIが当たり前の世代に、学びと仕事と未来を切り分けない場所をつくりたい。MIXIはAI企業として、伝えられることは沢山伝えていきたい。生徒も、我々大人も、より変化していけるとおもいます。 MIXIとRePlayce、今春開校した「HR高等学院 渋谷校」で「"キャンパスをつくろう" プロジェクト」を始動 prtimes.jp/main/html/rd/p… via @PRTIMES_JP

就労支援も素敵ですなー。AI時代なので今色々とまくれるチャンスなので、何か1mm でもと動き始めているところ。

MIXIデザイン本部、昨年度のクリエイティブをまとめたショーリールができました✨ AIならではの表現を模索しつつ、バーチャルプロダクション配信、ドローンショー、動画制作、WEBサイト、イベント体験設計と幅広く手がけてきました。 事業領域もモンスト、コトダマン、みてね、千葉ジェッツ、FC東京、TIPSTARと多岐にわたります。 「デザインする場」を広げる組織に、AIが自然とインテグレートされていっています。ぜひご覧ください。ナレッジや実例もだいぶ溜まってきたので、どこかでしっかり紹介できれば嬉しいなと思っています！ MIXI DESIGN SHOWREEL FY2026 youtu.be/vLoR86K4v8w?si…

神山まるごと高専に向けて話してくださってる？！と思うくらい学内で議論されていることや学生たちに伝わって欲しいことばかりで、驚きと感動です🥹✨ 関係者、必見！！！

動画に出させていただきまーした！ 自分にできることがあればと年間の日数を決めて、教育関連は動いているけれど、今最も産学でギャップがあると思うので、どうなるかわからない未来を沢山模索できるといいですよね〜。

A clip from a long-form interview I did this week keeps replaying in my head. We were talking about why AI hasn't only made work faster. It's also cracked open how people can learn. Until recently, absorbing a lot of information meant being a strong reader. Long PDFs and dense technical documents. Books were the bottleneck. People who couldn't sit down with a 300-page book were quietly disadvantaged. That's no longer true. The same source material can now become a slide deck, a webpage, a short video walkthrough, or an annotated diagram set for visual learners. For auditory learners, a podcast version. For someone short on time, a 5-minute summary. For someone going deep, a 30-minute breakdown that doesn't skip the hard parts. NotebookLM does any of these in a few clicks. Claude can reframe the same idea ten different ways depending on what you ask for. Learning works in reverse now. You see the finished output first, maybe as a podcast or a video, and then pull yourself toward the depth you actually want. Not "read 300 pages and maybe understand." The thing I keep landing on: learning has stopped being a pipeline issue. It's a routing issue. Match the format to the brain receiving it. I'm someone who absorbs through audio. A 30-page spec turns into a podcast and I listen on a walk, and I come back understanding it better than if I'd skimmed the doc at my desk. People I work with print things out. Some sketch over slide decks. We used to all pretend we were doing the same thing. The part that excites me isn't the model getting smarter. It's that "how I learn best" is no longer a constraint. The same content can meet me where I am. We're nowhere near the ceiling on this.

【#Newbee 2026/4/28公開］ ✅ あのMIXIは、もう全社AI導入を完了していた ✅ そして次なる山は「非エンジニア、学生教育」 ✅ MIXIは今何に取り組んでいるのか 取締役上級執行役員 @tatsuma_muさんに聞きました。 👇配信URLはスレッドに