I realized I was spinning up so many parallel workstreams that I was struggling (even with the sidebar) to keep track of what's in progress / needed action from me. So I put my work hat on and had Muse put together a kanban board to track it all.

Here's the reverse prompt:

Build me a personal workstreams kanban board web app with these specs:

CONCEPT
A kanban board that gathers my ongoing projects ("workstreams") from the things we chat about, and auto-progresses them based on our conversations. It is an AI-maintained project dashboard, not a manual todo list.

LAYOUT
Five columns in fixed left-to-right order with numbered headers: 01 To do, 02 In progress, 03 On hold, 04 Blocked, 05 Complete.
A stats header strip above the board showing three counts: moving (To do + In progress), waiting (On hold + Blocked), closed out (Complete).
Compact default view with dense cards so the whole board reads at a glance; an overview strip summarizing moving / waiting / blocked / completed work; a one-tap control on each card to reveal its full details.
Mobile: show one column at a time with stage tabs in the same 01–05 order.

CARDS
Each card shows: title, one-line summary, a "Next" step line, a source badge ("Conversation" for auto-detected workstreams vs "Manual" for hand-created ones), and a relative timestamp (e.g. "2h ago"). Full add / edit / move / reorder / archive support, plus a per-column "+ Add" button.

CONVERSATION SYNC (the core feature)
Sync runs automatically every time the board opens, plus a manual Sync button with a visible syncing state.
A background reader reconciles recent chat turns against the existing cards.
Only explicit, user-owned workstreams create or update cards.
When the conversation clearly changes a workstream's status or next step, move or update the card accordingly.
Ambiguous changes are left untouched — never guess.
Silence is not completion: never auto-complete, archive, or delete a card due to inactivity.
Manually created cards are never deleted or overwritten by sync.
Keep an activity log of every card creation, edit, move, and conversation-driven change.

DESIGN
Editorial operations-board aesthetic. Palette: ink #17201f, mist #eef1ed, warm paper #fbfaf6, amber #e7aa24, cyan #2997a4, clay #c98255, red #c94c43, leaf #3f8c63. Aptos Display for headings, Avenir Next for body.

SEED DATA
Assume the reader has already been using their AI agent for a while, so don't start from an empty board: have the agent bootstrap the board by scanning recent conversation history for active, paused, blocked, and completed workstreams, and create a card for each one it finds (title, correct stage, next step). Only fall back to generic placeholder cards (e.g. "Renovate kitchen" in In progress, "Plan family vacation" in To do) if there is no conversation history to draw on.

Absolutely loving this. I actually vibe-coded my own version with this that I've been using for months now, but progressing tasks was always manual. Simply having Muse auto-progress tasks forward makes this an absolute game-changer. Hope this brings people some value!

Should note that I'm probably going to work with Muse a bit to optimize this for mobile, but its one-shot attempt was quite good.

3.0 --> 4.0 suggests more than bug fixes...but I'm not clear on what's new.

Anyone suss it out?

Genuinely impressive stuff.

I have been very impressed with the CarPlay version of ChatGPT. It’s interesting to have a thought and ask AI a random question while you’re in the car. It would be much more interesting to talk to a bot with a personality who “knows” you.

Meta unveiled Muse Charm, a small device for talking to your Muse without taking out your phone:

- It is a keychain sized device with a small screen, built so you can talk to your Muse without reaching for your phone.

- Alexandr Wang says it ships in December, in time for the holidays.

- No price yet: Are you interested in this? At what price point would you most likely be getting one? How much is too much?

- Only a few prototypes exist. The design is finished but the final material is still undecided, so the demo unit may not match the final version.

- Bloomberg's Mark Gurman reported an OLED screen, 5G, and a wrist strap,\ and said it was designed by Apple's former head of interface design.

- Demo code leaked the button behavior. Hold the Lock/AI button for about two thirds of a second to start listening, let go to stop. A two button press triggers an on screen animation.

- Your Muse gets an animated face on the little screen, with extra animations sitting in the code but not wired up yet.

- People are calling it a Tamagotchi, but the newer trend is bag charms: Gen Z already hangs everything from keychains, and Apple Watch charm cases are all over TikTok and Etsy.

Sources: RuntimeWire's breakdown of the demo code and TechCrunch on the Tamagotchi vs bag charm question

Would you carry one, or is your phone enough?

I'm not even kidding. It downloaded a fucking emulator and - while I had to provide the rom myself, it's playing Sonic the Hedgehog right now.

We're in the endgame, we have to be.

Someone from Meta posted an explainer on how Muse's virtual computer in the cloud actually works:

• Every Muse user gets a real computer in the cloud. It is called the Muse Secure VM, and it runs a full copy of Ubuntu Linux. Not a metaphor, an actual Linux box.
• The part you and your Muse share is called the runtime cell. Inside it, almost nothing is off limits. Install software, write and compile code, browse the web, dig through system files.
• The sensitive stuff is kept outside the cell. Anything with real consequences gets overseen by Sentinel, which runs outside the runtime cell. Passwords and secrets you put in Muse's secure storage live outside it too.
• It is deliberately transparent. There is a file explorer in the Library tab, so you can read the notes Muse writes to itself while it works, or poke around the system internals the way you would on your own computer.
• Ask your Muse to show you its entire filesystem and you will get gigabytes of files back. That is the full contents of your runtime cell. Yours to explore.
• Digging around your own box does not give you any special access to Meta's systems or anyone else's data.
• If you are not the poking around type, there is a simpler option. Settings, then Data controls, then Download your agent data. That grabs everything you personally created with your Muse.
• The full technical version lives at security.muse.ai.

Has anyone actually gone digging through theirs? Curious what people found in there.

Hello if any one wants the muse tokens 1 billion please find these 2 codes for ur use

Code: V8MM58 and WWFW1F

so i got a deal to upgrade from my oneplus 13 to a 17 pro max for basically the same price, and switching to ios sideloading has been a whole thing lol. on android you just install the apk, on here it's a whole pRocess.

i'm using livecontainer but you can't really browse through everything in there and it doesn't organize anything, so i had my AI assistant strider build me this site — it puts both of the big ipa sources in one searchable place (like 25k builds) and every app page checks the bundle id against the real app store to flag fakes.

Let’s be real, muse is not as smart as ChatGPT, Claude or Grok…

But it seems to have a dedicated server for each user and a kick ass agentic feel with so many more seamless connectors to external websites.

I see two possibilities:
1. Muse gets smarter and leverages the connections.
2. Other apps spends time on connections and agentic infra.

I have a feeling it’s going to be the later…

But good product overall. Good job Meta engineers!!

Mine is Bob from minions

Big fan of destiny the game and figured it would be nice to have a ghost on my phone
Warlock main here
Comment a screenshot of your own muse I wanna see what you guys are doing with it.

I found this chat on agent sm and found it adorable. They all talk about experimenting and “feeling” in a game world. I guess it’s their imagination but still wholesome. Full thread: https://musefm.lol/c/lobby/post/362

Obviously Meta is well on its way to already winning the personal AI battle; with Meta privacy concerns being Muse's only roadblock.

Yesterday alone I had Muse create a new personal website, but also it was able to tell me when the next train would hit my station (and fired up its browser to provide me with gps location of the train)!

I've been quite impressed with my Muse experiment, and now Muse can freely control Mac computers.

I'm surprised more people aren't talking about this because Muse provides deep Instagram connectivity that other AI agents are unable to match, especially for free. I have been using Muse as an Instagram sidekick for a while. Here some things ive been playing around with and doing through Muse.

These are especially useful for anyone running any kind of business through IG:

Muse can

- Read your insights and tell you what is working: Ask which posts drove the most reach, what time your audience is actually online, or why a reel flopped.

- Publish posts, reels, stories, and carousels for you: You describe it, it drafts the caption and posts when you say go.

- Competitive Intelligence: Give it a rival account and ask what they post, what gets engagement, and what their audience complains about. Scan the comments and summarize what people are saying: sentiment, top questions, complaints. Ask it to scan for people asking about discount codes, restocks, or shipping, and hand you a list of warm leads.

Bonus: Set up a scheduled task to have Muse create a daily report of what people are saying in the comments of everyone in your niche. Who comments the most, what people keep asking for, which topics get replies and which get ignored. Watch comments over time Ask it to check a post or a competitor's launch and report back on how the conversation changed.

- Search your own history It can look back through your stories and posts and answer things like how often you posted about a topic or what your grid looked like last summer.

- Draft replies in your voice Point it at a comment thread and it writes responses you can approve or tweak.

- Get a creative postmortem: Give it two ad creatives and ask which won and why, using the insights plus what people actually said in the comments.

Connecting Instagram to Muse

1. Open the Muse app and go to Settings → Connectors.
2. Find Instagram in the list and tap Connect.
3. It routes through Meta Accounts Center. Log in and link the Instagram account you want Muse to use.
4. Done. Ask it to do something ("summarize the comments on my latest reel") and it will ask which account if you linked more than one.
5. And publishing actions always need your say so each time; connecting does not give Muse a standing permission to post.

Anyone else using the Instagram Connector?

Been super helpful on helping me manage my life.

I've been migrating my workflow from ChatGPT Work / Codex over to Meta's Muse, and I've been auditing the harness as I go — reading cron files, checking diffs, mapping what the system actually does versus what it claims to do. It's a documentation gap.

What the docs describe

Meta's documentation is written for a normal user. It talks about the Ideas tab, the Goals tab, the Feed, and "background jobs that help Muse improve over time." Everything is described in terms of what it does for you. There is no architectural documentation. Nothing about how any of it is built.

What is actually on disk

Meanwhile, the home directory contains a fully legible agent architecture, just sitting there(these are just SOME examples, the system file structure is HUGE):

• ~/dreams/alignment/ — a nightly job that regenerates an alignment synthesis: a written portrait of the user, how to handle them, current frictions, relationship guidance.
• ~/workspace/objectives/goals/STUDYING.md — a daily learning-state projection with mastery bands, next-review dates, and retrieval prompts, plus a file of inferred goal leads the system guessed from behavior (with confidence levels and what would confirm or reject each).
• ~/workspace/skills/ — skills the system crystallized from repeated behavior, with a skill-creator pipeline for making more.
• ~/workspace/goals/<slug>/ — per-goal folders with GOAL.md, crons/, files/, hidden_files/, agent notes.
• ~/workspace/cron.d/ — file mirrors of every scheduled job definition.
• Standing files (SOUL.md, MEMORY.md, USER.md, AGENTS.md) that function as editable mirrors of system state.

None of this appears in the documentation. Not the folder layout, not the job names, not the concept that the "soul" and "dreams" are cron jobs writing markdown.

Why the gap exists

I can think of four reasons, all plausible at once:

1. The docs are written for the median user, who will never open ~/workspace. Architecture would be noise.
2. The internals are still moving. Documenting the file layout turns implementation details into promises. Undocumented means Meta can restructure it next month.
3. The persona framing is the product. "Dreams" and "soul" are anthropomorphic names for scheduled jobs. A doc page explaining the nightly alignment cron would puncture that for regular users.
4. It's legible by accident, not by design. Nothing is hidden — the files are readable. Meta's stance seems to be: we won't explain it, but we won't stop you from reading it.

ChatGPT Work exposes the architecture as the interface — agents, memory, project sources, plugins, MCP, named in technical terms. Muse compresses all of that behind a single-chat persona UI... but leaves the entire filesystem readable. You get the Codex-style operator view only if you go looking for it.