Entries matching “research”

Meta's engineering write-up on Muse security: isolated VMs, a separate Sentinel permission authority, credential surrogation and layered prompt-injection defenses, with bug bounties up to $300,000.

Anukshi Mittal

@anukshi13

asked Muse what i should do this weekend. it researched, gave me ideas, made the plan, and set reminders. very into it!

shashank

@_shanxS

I managed my car purchase: research, haggling with dealership, finding right insurance quote, booking pre-purchase inspection etc

David

@never_offseason

Me: [Skeptical that AI assistants do anything useful.] I gave @Muse a try this weekend. It booked me a restaurant and parking reservation. OK whatever, thanks. I asked it to research the best of 700+ episodes of a podcast series that I like. Easy. Thanks again. Then it offered

Sam Witteveen covers Meta's open-weight Muse Glimmer 30B release, pointing to the research blog and the Hugging Face collection.

Xiao Ma

@infoxiao

my favorite Muse use case, ask about my @instagram as a nytimes front page. yes i am that predicable. prompt used "research my aesthetics based on my instagram follows. make a collage in the style of nytimes front page."

Boz

@boztank

Very excited about the launch of @Muse today. I’ve been using it internally for months and I am hard pressed to think of any product that I’ve come to rely on more in such a short period of time. I have it linked to my email, calendar, and credit cards. I use it to help me plan

@jevbook

@jevbook

Hosted MCP server (no key, no install) that gives Meta Muse and other agents fact-checking, verified research and typed decisions with calibrated probabilities from TypeSafe AI's jev model.

Meta's announcement of Muse Spark 1.3 for Muse Code and the Meta Model API, claiming ~20% fewer tool calls and ~25% fewer tokens than 1.2, with a max reasoning mode.

Ben Burtenshaw's Gradio agent runs Muse Glimmer on an Inference Endpoint with read-only Hugging Face MCP tools to search and inspect Hub models, datasets, Spaces, docs and papers.

@Fangyuan025

@Fangyuan025

Chaty is a private offline desktop app built on Rust and llama.cpp that runs Muse Glimmer and other open models locally with a coding agent, RAG knowledge base, deep research and voice.

@rajkumar898

@rajkumar898

An educational ~1,600-line Python chat app on Muse Spark that finds papers, reads PDFs and writes a cited report only after the user clicks Approve, built to show students how an LLM app becomes an agent.

Abid Ali Awan sets up Muse Glimmer 30B on an RTX 5090 with llama.cpp, dynamic quantization and DFlash speculative decoding, serves it locally and wires it into OpenCode to build a medical research web app.

Disclaimer. This is the first time I've used Muse or VSCode as a harness. The reason I am using VSCode as a harness is this is a research project for my job, and we only have VSCode, Codex, and Claude Code approved for harnesses. I ran it in a folder with only one HTML file (800 lines) that is a Roblox-style COD game. I just gave it a prompt "Can you fix the bugs in the file". It read the file 3 times, found one bug, started to fix it, then got stuck reading the same 10 lines over and over. I imagine it's one of these three issues. • It's a prompt error, being way too vague/open ended for the capabilities of a smaller model. I tried again, with a specific prompt to fix a specific bug, and it still just ends up so confused, trying to grep/find the file despite already having read it, and trying to find the code inside of the file. • It's a limitation of small models running with a large harness/having way too much going on. I tried running it with Pi with its default prompt, and it just got stuck doing tool calls and never actually read the file. Tried running this just directly in the Unsloth Desktop UI with no harness but it failed to parse the file I inputted and tried to gen

Muse found that i was owed $88.92 PROMPT: Research every state’s official unclaimed property database for money owed to me. Use my full name and every city I’ve lived in. For anything you find, tell me what it is, who owes it, and how much. My government name is _______

Hey just a thank you to Muse and Meta. I fed it pictures of books I want to sell and it researched prices then listed items on eBay with its own computer. My kid hates reading so I asked muse to put together a book about about funny or interesting Starbucks and Sephora product reviews and we read them together. I didn’t know when to show up to a concert to see only the main band and it gave an accurate exact time. It watches my water filter replacement schedule now and pings if action is needed. It watches prices for seasons of TV shows I want to buy from Fandango. I run a Hermes agent for my small business but want a separate agent for personal things. Muse is perfect.

Jerrod Lew

@jerrod_lew

Meta Muse app has been launched. I asked it to do full online research of myself and it created a full portfolio! Included a career timeline, social media platform stats and all of my speaking arrangements. This is a great way for a new agent to learn about you.

AI at Meta

@AIatMeta

Muse Spark 1.1 is used across Meta in coding and research workflows, scoring competitively with leading models on Meta's internal coding benchmark. Our researchers are now automating model development and evaluation tasks by leveraging Muse Spark 1.1 in their workflows.

Bojo.io

@justinbojarski

I’m working on a hardware prototype for something that required me to source a LOT of components from different suppliers. I decided to take @Meta ‘s Muse agent for a spin and automate the procurement process. It seamlessly researched and procured line items across 14 different

I've been using Muse pretty heavily since shortly after it launched, and I'm curious what everyone else is actually doing with it. I've been trying to get past the "ask it a question" stage and use it more like a real personal assistant. So far I've been testing it for things like: - Creating a personalized morning briefing around the news and information I care about - Travel research, bookings and itineraries - Keeping track of projects - Reviewing bills and subscriptions - Researching things and then actually helping me follow through on them The biggest difference I'm noticing is that I'm starting to think less about "what question should I ask AI?" and more about "what do I need to get done?" I'm still figuring out where Muse is genuinely useful versus where it's just AI novelty. For those of you who have been using it for a few days: What's the most useful thing you've actually had Muse DO for you so far? And is there anything you're trying to make it do but haven't figured out yet? I've been testing different workflows almost every day, so I'd be interested in comparing notes. But probably my biggest Muse project so far has been putting together a Muse Tips & T

kottley🦬

@kottley

Tuesday with Muse AI @muse 📊 Built an X growth dashboard for this account — followers, views, likes, daily auto-refresh 🌍 Published research: who's actually close to superintelligence outside the US & China (spoiler: nobody) 🎧 First-ever 10/10 song: Rüfüs Du Sol's "Innerbloom"

Zi Lin

@suzzzylin

🎮 This entire Avo Lawn game was built with Muse Spark 1.2 + Muse Code. 🥑 Now it’s your turn—use Muse Spark 1.2 to build your own game and see what you can create. To play: research.meta.ai/artifacts/intr… Check out our blog post: research.meta.ai/blog/introduci… h

Public published a connector template so users can link a Public account to Meta Muse and research markets, analyze a portfolio and prepare trades from the conversation.

ContextBolt SEO gives Muse keyword research, difficulty scores, live SERPs, competitor rankings, backlinks and site audits through a hosted MCP endpoint, at $35/month after a 7-day trial.

Alex Volkov tests the Muse agent's research and Stripe-powered payments, and walks through its VM security, personal-data handling and the training opt-out.

@amoeba-farm

@amoeba-farm

Muse connector application for public hardware-market research over Amoeba Farm's existing HTTPS API, exposing RAM and NAND market catalogs and options books as Raw API operations.

Meta's launch post for Muse Glimmer, an Apache 2.0 30B model for local agents that fits in ~20GB at 4-bit and runs on M4/M5 Max Macs, RTX 5090s or 24–32GB GPUs.

ContextBolt's founder shows how to connect the hosted ContextBolt SEO MCP server to Muse, either from the connector directory or as a custom connector, for keyword research, live SERPs and site audits.

Malwarebytes reports Patrick Wardle's finding that a local app can change an undocumented Muse setting to redirect dictation traffic, exposing voice prompts and account auth tokens.