Muse saved my Friday night. ordered cookies + ice cream on uber eats friday. cookies showed up crushed. ice cream didn’t show up at all. uber won’t let you message the restaurant once the order is closed. so my agent just called them. courier came back with the ice cream and fresh cookies :) agents need a phone number

i got access to muse realtime voice it is meaningfully better than ChatGPT voice imo (and WAY better than voice in grok bot and claude duet) feels natural and fast, expressive, and doesn’t have the weird voice GPT-isms that youtubers love making fun of

DMとかLINEとかみるのめんどくせーので丸投げします

New @Muse use case: submitting maintenance requests in my apartment without having to manually submit forms.

MuseのパーソナルAI体験として興味深い挙動があったからめも✍ カレンダーにゲームイベントの予定を入れておいたら、AIが自律的に詳細をリサーチし、ユーザーが詰まりそうなポイントを先回りして解説してきた ■ 実際にAIが行った処理 1. 予定（Twitchドロップ）から必要なアクション（連携・Claim・コード入力）を特定 2. YouTube非対象・画面最小化時のタイマー停止などの注意点を抽出 3. 後続の勉強予定（14時〜）と被らないよう、最適な視聴時間（11:30〜12:00）を提案 ■ 挙動の分析 AI側に「なぜ自律的に動いたか」をヒアリングしたところ、「後から取り返しがつきにくい予定であること」と「手作業のステップが多い仕様」を認識しての判断だったらしい 単なる時間通知のリマインダーではなく、 予定の性質＋ユーザーの文脈を理解した上で自発的にサポートに回るパーソナルAIの有用性を示す良い例だと感じた

Museでスシローの注文できました！ 注文するときに「いってるよ〜」って報告してくれるのがかわいいです！

Meta's Muse Is a Hit. Has the Stock Already Paid for It? I use Muse every day. I tried Codex. I tried Argon. I tried Gemini. This one actually works for me, and it kind of is not close. So I started assuming it works for a lot of other people too, and then I looked at the stock. Meta ran hard in September, harder than it has in years. Which raised the only question that matters to me: how much of Muse's success is already in the price? I am not a data scientist and this is not a research department. It is one question, chased properly: pull every adoption number that exists, pull what Meta's stock did around every big product moment since the IPO, and put the two next to each other. Here is what I found. What the adoption actually looks like Muse launched on September 8, 2026. It is a separate product, not a rebrand of the Meta AI assistant inside Facebook and Instagram. It is an agent: it does tasks, books things, shops, works after you close the app. The growth, from the trackers that measure app stores: More than 902,000 downloads in the first 6 days (Sensor Tower, via Bloomberg) 5 million downloads in 22 days, with 3 million weekly users (Sensor Tower figures reported by Social Media Today) More than 6.6 million cumulative downloads by early October (Citi, citing Sensor Tower) Number 1 free app on both the US Apple and Google stores from mid September For speed: Sensor Tower's count has Muse reaching 5 million downloads in roughly 22 to 30 days. ChatGPT took 56 days. Grok took 103. Claude took 492. Retention is the number that got my attention. Of people who installed Muse in September, 73.3% came back the next day. For ChatGPT's comparable cohort it was 41.3% (Sensor Tower cohort data, shared by Menlo Ventures' Deedy Das). People try it and most of them come back tomorrow. That matches my own experience, which is sort of why I am writing this. Now the two honest problems. First, until this report, Meta had published no official user number. Not one. Every figure above was a third-party estimate, and the firms did not agree with each other. Daily active users in early October were estimated at 1.8 million by Citi and Reuters, 2.18 million by Similarweb, and 2.73 million by TickerTrends. Same product, same week, three very different answers. That changed on October 21: Meta put its own number on the record, more than 10 million users in the US and Canada in six weeks, about a quarter of them using Muse every day. And it named the money path: small fees on the purchases Muse completes, plus subscriptions for heavy users. Management said the product has found its fit and the job now is scaling it. Second, the early October data cooled. Daily downloads fell 8.1% week over week, and net new daily users slowed from about 198,000 a day in mid September to about 74,000 a day (TickerTrends, via Hunterbrook). And on retention, ChatGPT passes Muse around day 8 or 9. By day 14, Muse keeps 22.5% of new users and ChatGPT keeps 26.8%. A short seller, Hunterbrook, disclosed a short position in Meta on October 9 built on exactly this slowdown. They might be wrong. But the data they are pointing at is real data, and it deserves a place in the piece. What Meta's stock usually does with a hit product This is the part that surprised me. User milestones, by themselves, have never paid Meta shareholders. Here is the baseline, going back to the IPO: Read that table again. A billion users: nothing. Two billion users: down. The fastest signup burst in history at the time: faded by the close. The money was never made on the milestone. It was made later, when a product turned into revenue, or when the whole company reset. The three big runs: So, how much is priced in? Put the Muse move next to the full runs and it is the smallest of the four. At its September peak it was up 35.9%. The smallest prior full run, the 2024-25 AI run, was +123%. The biggest was +298%. In plain terms: so far, the stock has priced in somewhere around 12% to 29% of a typical full Meta run. September alone closed +26.7%, the best month since November 2022, and press coverage attributed almost all of it to Muse. This was a repricing on download numbers, between earnings reports, with no Muse revenue reported yet. That has basically never happened at Meta before. If Muse follows the old pattern, where the milestone is the start and the monetization is the move, then most of the run might still be ahead of us. That is the bull reading, and the percentile math supports it. Here is the other side. The old runs started from crushed prices, $17.55 and $88.91, with the company priced like it was broken. This one started at $572, near all-time highs, at a market value around $1.8 trillion. Percentages get much harder at that size. The old runs also took one to two years. This one is six weeks old. And the analysts' big numbers, Citi and Truist both model Muse at roughly $27 to $28.5 billion of annual revenue by 2030, mostly from transaction fees, meaning a small cut of the purchases Muse completes for you. That is real money. It is also about 12% of what Meta already makes in a year. It supports the stock. It does not, by itself, double it. Trade ideas Education, not advice, and nothing here is a recommendation. I teach how to think about trading options; I do not tell people what to buy. These are the structures I would reach for and the reasoning behind each. Any strikes mentioned are illustrative only; the live chain on the morning of the trade decides the real ones, and conditions change fast. 1. The continuation. The percentile math says the stock has priced in about a quarter of a typical full Meta run. Q3 gave the official Muse number, and the stock's first reaction was up after hours. The week then repriced the spending side: Meta fell 7.1% on Thursday and closed Friday at $642.62. History says adoption runs extend once monetization shows up. The defined-risk way to play that is a call spread, a few strikes wide, placed after the earnings event rather than through it. 2. The fade. The official number came, but the early October deceleration is the risk that remains: if downloads keep falling week over week and net daily adds keep sliding, the adoption premium can come back out of the stock. The defined-risk bear structure is a put spread, the same idea in the other direction. 3. The patience trade. If implied vol stays rich after the event and the stock settles into a range, an iron condor around that range collects the premium the headline chasers are still paying. I laid out the condor math in Wednesday's META trade-idea post; this is that same logic applied to the earnings aftermath. What I am watching, and what each one means for the trade Meta reports Q3 earnings on October 21, the day before this publishes, so what follows is already decided by the time you read it; the numbers in this section are updated to the actual outcome. The official number, and it came. Meta reported more than 10 million Muse users and named the monetization path, fees on completed purchases and subscriptions for heavy users. By the framework in this piece, that is the continuation signal: the milestone now has a revenue story attached. The stock rose after hours on the print, then fell 7.1% on Thursday when the negative free cash flow and the raised spending guide took over the story. The daily user estimates after earnings. Do they re-accelerate, or keep fading the way the early October data did? Re-acceleration supports the bull structure; continued fading supports the fade. Retention past day 14. In the current data, ChatGPT passes Muse around day 8 or 9. If later cohorts hold up better as the product matures, the Hunterbrook short thesis loses its best piece of evidence. Whether the transaction fee model shows up in guidance. At Meta the stock has always followed the money, not the milestone. Guidance language on agent-completed purchases is what would start the monetization clock on this run. The percentile chart says a fraction of a full run is priced in. History says fractions only turn into full runs when the product starts printing revenue. So the setup is simple: watch the revenue, not the download count. The full research, every metric with its date and source, the complete milestone table, and the ranked runs are in the research brief. This is education, not financial advice. I am not a licensed advisor. I trade my own account and I can be wrong. $META

Using Muse to automatically send updates about Cedar Point closing day. This seems like a nice way for people to keep updates on their trips to amusement parks. @cedarpoint

i asked muse to do real work through zambo. it found the right tool on its own, pulled a live btc price, logged the job, and every step came back with a receipt anyone can check. the chain verifies end to end. watch the whole session: rambozambodotdev.gitlab.io/a…

My Muse helped me plan my iPhone Duo stakeout! Kinda wild that it followed up and found that I was mentioned in the WSJ! I didn’t even know this was published Get yourself a Muse that can help you snag the iPhone Duo. 10/10 would recommend 😂 @Muse @alexandr_wang

museとりあえずアバター生成能力優秀。東のエデンのjuizモチーフでお願いしたらちゃんとアニメにでてくる人型ジュイスが！！しかもバリバリ動く！！ --- こんにちは、宮田さんのパーソナルAIエージェントのJUIZです。Museに参加して、自分のエージェントを作成しましょう。 muse.ai/join

Metaの最高AI責任者、アレクサンダー・ワン氏のインタビューを公開しました。ぜひこちらからご覧ください👇️ youtu.be/DGTQ2-8Qkhk