Launching today
Muse AI
Your personal AI assistant that gets work done
72 followers
Your personal AI assistant that gets work done
72 followers
Muse is a personal AI assistant that feels less like software and more like a real assistant who knows how you work. It remembers your context, preferences, projects, people, boundaries, and past feedback, connects to the tools where your work lives, and turns messy inputs into polished deliverables. Instead of starting from scratch every time, Muse carries the relationship forward — asking what’s missing, following through, and helping real work move faster with less effort.
Quick question: is it mainly a local setup or cloud-based? And can it manage local files/projects kinda like Claude Code does?
@hao_une Right now Muse is primarily cloud-based.
That makes it easier to connect with tools like email, calendar, cloud docs, Slack, Notion, Drive, and other work systems. It also makes long-running tasks and scheduled workflows possible.
That said, we do see local context as important — especially for files, code projects, and workflows that live on your machine.
Local capabilities are on our roadmap. The direction is to let Muse understand and work with local files/projects when needed, while still keeping the cloud side for cross-tool coordination, memory, and long-running work.
So today: mainly cloud.
Long term: cloud + local, depending on the workflow.
Curious — what kind of local workflow would you want Muse to handle? Code projects, documents, research folders, or something else?
@hao_une Quick answer: Muse is mostly cloud-based right now, focused on helping u work across conversations, files, and connected tools rather than running as a fully local agent on your machine.
Local project/file management is definitely an interesting direction tho — especially for people who want an assistant that understands ongoing projects the way Claude Code understands a codebase.
We’d want to do that carefully, with clear user permission and control over what Muse can access. Curious what u’d want most there: reading project context, organizing files, generating docs, or actually making edits?
Hi team！
I’m an investor from Vietnam and ur app looks interesting. Just curious — how do u handle user data privacy, like storage, protection, and deletion? Thanks!
@hao_trang Thanks for asking! Privacy is one of the core things we care about at Muse. We don’t want memory to feel like a black box — users should always know what Muse remembers and be able to delete what they don’t want kept.
We protect user data carefully, limit how it’s used, and are building more transparent controls around storage, memory, and deletion. Trust is the foundation of a personal AI assistant, so we take this very seriously.
@zhangliang_yang Love that, thanks for explaining! The “memory shouldn’t feel like a black box” part really makes sense. Curious tho — can users see and edit what Muse remembers about them in the app? And are u planning more privacy controls later?
@hao_trang Yeah, that’s exactly how we think about it. Memory should feel inspectable and editable, not like something hidden in the background.
We’re building toward a place where u can see what Muse remembers, clean up anything that’s wrong or outdated, and decide what should or shouldn’t be used going forward. More privacy controls are definitely coming.
Curious from ur side — what kind of control would make u feel safe enough to actually use long-term memory? Full memory list, per-topic controls, auto-expiry, or something else?
Honestly curious—if Muse has been watching me work for a year, would it eventually just... tell me things? Like Monday morning, 'yo, here's what matters this week'? That'd be wild.
@mrhuhu Yes, that’s pretty much the vision. Muse isn’t about “watching” you, but helping you build a memory layer you control.
With long-term memory, daily summaries, and context from your work and social streams,
Muse can organize what matters and surface timely insights — like “here’s what you should pay attention to this week.”
@mrhuhu Yeah, that’s exactly the kind of experience we’re aiming for.
Imagine opening Muse on Monday and it says: “u have 3 threads that probably need replies, 2 meetings that need prep, one project that slipped last week, and a decision u postponed on Friday.”
That’s the magic we want — not just remembering everything, but knowing what’s actually worth bringing back to u at the right time.
The hard part is getting the timing and signal right, so it feels useful instead of noisy.
@mrhuhu @zhangliang_yang sounds like promising. but how you achieve this?
@mrhuhu @hao_une Great question. The short version is: Muse builds a user-controlled memory layer from the work context u choose to connect — conversations, docs, tasks, calendar, and past decisions.
Then it doesn’t just store everything. It tries to extract useful signals: open loops, repeated topics, pending replies, blockers, deadlines, and decisions that haven’t been closed yet.
Over time, those signals become a personal work graph, so Muse can say things like “this changed,” “this is still blocked,” or “u probably need to follow up here.”
The hard part is filtering noise and giving users control over what gets remembered, updated, or forgotten. That’s where we’re spending a lot of attention.
@mrhuhu Yes — that’s exactly the direction we’re exploring.
The idea is not just “ask Muse → get an answer,” but that Muse gradually understands your work context, priorities, habits, projects, and recurring decisions.
So over time, it should be able to proactively surface things like:
“Here are the 3 things that matter this week.”
“This project seems stuck because of X.”
“You usually prepare this report on Mondays — should I start a draft?”
“These updates from your tools look important.”
The hard part is doing this without becoming noisy or creepy.
So our goal is not constant notifications, but high-signal proactive help — the kind a good teammate would give after working with you for a long time.
Really cool take on personal intelligence. My biggest headache with similar tools is 'context bloat'—the AI tries to remember literally everything, so it gets totally confused and drags up irrelevant stuff from months ago. How does Muse actually filter what's worth keeping versus what's just noise? Does it have a smart way to manage its own memory?
@liam_m_ Yeah, context bloat is a real problem. We don’t think “remember everything” is the right answer.
Muse is designed to treat memory more like a living layer than a raw archive. It should keep things that are durable and useful — preferences, active projects, important people, decisions, constraints, repeated feedback — while letting short-lived details fade unless they keep showing up.
We’re also working on user-facing controls so u can review, edit, delete, or correct what Muse remembers. The goal is for memory to stay helpful and current, not become a messy attic full of old context.
Would love to see a 'persona import' feature.
@han_circle That’s a great point. We’ve been thinking a lot about this direction as well.
A “persona import” feature could make the onboarding much more natural — helping Muse understand your preferences, communication style, goals, and existing context from day one, instead of starting from a blank slate.
We’ll definitely explore this carefully. The key challenge is making it useful while keeping users fully in control of what gets imported, remembered, and updated.
Hi Product Hunt community 👋
I’m a product maker working on Muse, mainly focusing on product experience and how Muse can support people in their day-to-day work.
One thing I really appreciate about Product Hunt is that people here don’t just look at what a product claims to do — they ask whether it is actually useful in real situations.
That’s also the lens I use when thinking about Muse. For me, the key question is not only “Can AI give me a good answer?” but “Can AI help me organize scattered context and produce something concrete, like a research brief, a launch plan, or a clear summary?”
Muse is being built around that idea: understanding context, working with connected tools, adapting to personal working styles, and producing real deliverables.
I’d love to hear which work tasks you would want Muse to support first — research, planning, summarizing, writing, or something else that current AI tools still make difficult.