got my mind on ai and ai on my mind. slinging tokens on @aiatmeta
USA
Joined July 2009
- Tweets5.4K
- Following865
- Followers19.2K
- Likes11.7K
Jack Krawczyk retweeted
I’m hiring an intern to work closely with me at the intersection of Health, RL and agents at MSL 🧑⚕️
Directions I am very excited about are taking shape, with plenty left to explore.
I’d love to hear from curious people who enjoy turning open-ended ideas into things that work!
agreed as a lower bound
Meta Muse will probably create 1000+ millionaires
This is the most slept on arbitrage I’ve seen in years and it’s not even close.
Here’s how i’d build a 1 person business powered by AI/Muse with with $0 and 0 social following
Jack Krawczyk retweeted
As promised, Muse Spark is being open sourced today.
This was our first base model we trained under the new MSL org, this open weight version has gone through a lot of careful additional training for open-weight release. It’s an efficient and capable model, try it out!
Jack Krawczyk retweeted
Muse Spark Open -- weights releasing today -- runs great on new Mac Studios thanks to work with @johnternus and Apple.
the only thing better than Muse Wonder Full is updated token efficiency in Muse Code to resultmaxx
Muse Code 1.4.8 is out and is focused on token efficiency.
- Auto-review and background agents now use a cheaper model when available, and long auto-reviews stop sooner instead of burning tokens.
- We've fixed a bug that routed a small number of users who had enabled contributor mode to the full-priced models.
Jack Krawczyk retweeted
muse wonder is super fast and token efficient ⚡️ ⚡️
give it a go!
The two top models on the Intelligence Index take about twice as long per task: Claude Opus 5.5 and Sonnet 5.5 (@AnthropicAI) vs. Muse Wonder (@Meta).
Muse Wonder finishes an average task in 6.8 minutes with a 52.4 score. Opus 5.5 takes 13.2 and Sonnet 5.5 takes 15.3, scoring 57.6 and 56.0. Muse Spark 1.3 and GPT-6 Astra (@OpenAI) follow at 8.1.
When someone is waiting on the answer, minutes can matter more than a few index points. Merge Gateway lets you send latency-sensitive requests to the fast model and the hard ones to the strong one, behind one endpoint.
Scores via artificialanalysis.ai, week of Oct 12.
Jack Krawczyk retweeted
Build the future of superintelligence with us.
The Meta Global AI Developer Hackathon is taking place from Oct 26-Nov 4.
Apply to compete for a share of $1M in prizes and build with our newest models. Each selected participant gets $150 in Meta Model API credits.
Limited to 20,000 selected applicants. Terms apply → bit.ly/4saQrA8
Apply now → bit.ly/4hR3cuQ
Jack Krawczyk retweeted
If you ran into usage issues earlier today, this has now been fixed. We've reset everyone's usage as compensation. Happy coding!
Jack Krawczyk retweeted
Personal finance and legal are important capabilities for Muse, and Muse Wonder is extremely strong on them! 💪🍉
When you consider Muse Spark was April, Muse Spark 1.1 was July, 1.2 was August, 1.3 was September... it leaves you full of Wonder.
get the full experience: curl -fsSL dev.meta.ai/install.sh | bash
Jack Krawczyk retweeted
@Muse is now on iPad! Kudos @ivanbruel for sprinting on this!!
using @Muse can save you $2,000 or more on your car insurance 🦎
I put Muse, Claude and Gemini to the test to get quotes for Home and Auto insurance for me since I’m up for renewal soon.
Typically, I wouldn’t be inclined to shop since the hours of work leads to negligible difference in quotes. The juice was not worth the squeeze so to say.
So the inspiration here, was can AI lug through the sludge and if that saves me time and effort, in my opinion that would be worth saving a couple hundred bucks.
Gemini (Pro) was the worst and the most shallow. You’d think with integrations to my email and calendar it would go further, instead all I got was generic corporate speak.
Claude (Pro) was the next best, it read documents and screenshots from my current policy with Allstate and found some novel insights.
Muse (free) spun up 3 virtual computers, filled out the forms, asked me questions when it was stuck, and allowed me to take over use to fill in private information like SSN directly into the form.
Progressive’s quote completed first (in about an hour) but didn’t match the coverages I currently had so ultimately had to rerate after adjustments.
Liberty Mutual conked out after 5 retries and ultimately failed. No progress saved either.
Geico’s completed in about 2 hours after 1 retry.
Net result, it found $2,000 in annual savings without reducing coverages and limits from what I already have.
Looks like I’m switching back to Geico.
Note* - I’m a designer at Geico, this isn’t a paid endorsement or anything. Just my personal experience and I’m frankly surprised at the outcome (that this approach actually worked 😳, how long it took 😩 and how much savings it uncovered 💰).
Jack Krawczyk retweeted
🚨 side project alert 🚨
Announcing Muse Gadgets, an open source ESP32 firmware and Linux sdk so that you can make hardware devices that work with Muse.
Grab an API token from gadgets.muse.ai and point your favorite coding agent at the github repo to build your own peripherals for Muse.
Jack Krawczyk retweeted
We just crossed 3,000 connector submissions for @Muse Connector Platform! And today we're releasing a developer portal to make it easier for devs to complete intake, manage and track your submission. muse.ai/platform
Incredible stories I heard from business owners leading up to launching Muse for small business and how it created more time & opportunity for them:
1) Ad Optimization: Comparing sales volume in their Shopify store to external events tangential to their business, mapping it to ads spend, measuring ROAS during those external moments... then having Muse adjust their ad budgets to expand during demand surges dynamically.
2) DM Organizing: For every DM via Instagram over the past month (over 100k messages!), categorize the types of requests received then identify any missing business needs, inconsistencies, and even publish orders that came in via DM to their order fulfillment service.
3) Expanding Sales Footprint: A retail business sells by going to local markets and Muse actively searches for new markets within driving distance and shows them to the owner... then fills out the form to get a slip at the market.
4) Keeping Website Current: Based on inventory and new produced products, this owner messages his Muse about what's happening, sends pictures, and then Muse automatically updates the site with the information.
5) Employee Load Balancing: A local chain of a few locations plugged in their point of sale system and found that one location was seeing lower than normal sales and the other was higher than normal. The owner used that to send a few employees from the lower sales to higher sales location to help meet the higher demand in real time.
These stories keep building and continue to inspire creating the future for everyone.
We put this love into muse.ai/business - because nobody started their small business to focus on back office activity... they did it to engage their community with their passion.
We are moving from scarcity of time & expertise into an abundance of it so that everyone can spend more time with their community, creative process and teams.
Here's to building stronger businesses through more time back to focus on what matters most!
Deeply grateful to the entrepreneurs who helped us test it out before this week's launch. Read their story and support their business! nypost.com/2026/09/29/tech/m…
