
Leading Product @SpaceXAI (from @cursor_ai), previously Product @meta, @algolia & @zendesk, 🇫🇷 in CA
San Francisco
Joined April 2009
- Tweets7.9K
- Following3.3K
- Followers4.4K
- Likes3K
This is the coolest thing I’ve read here in a while. Props to all the very good doggos out there
Your dog spends about 20% of the night awake and its sleep quality doesn't drop at all. In the bed-sharing arrangement, you were always the fragile one.
Dogs are polyphasic sleepers. Their sleep comes in short bouts stacked across the whole day, and their brains are built to surface, check the room, and sink back under in seconds. Humans run on 90 minute cycles where a single bad 3am wake-up can cost an hour. Same mattress, two completely different operating systems.
A Mayo Clinic study put activity trackers on 40 owner-dog pairs for a week. The dogs hit 85% sleep efficiency whether they slept on the bed or the floor. The humans managed 81%, and their number moved depending on where the dog was. Dog sleep was bulletproof. Human sleep bent around the dog.
Then wearables blew up the sample sizes. A 2026 analysis of 11,733 WHOOP users across 1.6 million nights found that on dog-in-the-room nights, people slept about 10 minutes longer, resting heart rate dropped about 1 beat per minute, and heart rate variability rose. There were a few extra micro-wakeups, around 0.4 per night. The nervous system downshifted anyway. It read the dog as safety.
A 2025 study ran sleep EEG on family dogs and found the inversion. Dogs sleep measurably worse when their owners aren't in the room. You are your dog's sleep aid, and apparently its main sleep disruption too.
The evolutionary read is that this deal is tens of thousands of years old. Humans who slept near dogs could afford deeper, more vulnerable sleep because something with better hearing was handling the night watch. Co-sleeping researchers think dogs literally expanded how much uninterrupted sleep early humans could get.
The dog takes the broken sleep so you don't have to. The trade predates agriculture, and your heart rate still drops a beat the moment one curls up at your feet.
Maxime Prades retweeted
Two things matter most when hiring executives or promoting leaders internally:
1. Can they hire exceptional people?
2. Can they solve hard problems?
External executives are usually experienced at building teams, but often lack the context and content depth to solve the hardest problems themselves.
Internal leaders usually get promoted because they’re exceptional problem solvers. The question is whether they can become exceptional at hiring.
The best leaders can do both.
Love seeing this!
Starship has begun deploying its payload of 26 @Starlink V3 satellites. This deployment sequence will take ~30 minutes
This is such a great feature
SpaceXAI: Grok Bot now lets you route egress through your local machine.
(It means Grok Bot's VM will connect to things from your IP address, not the datacenter's.)
This will be great to access picky sites and services that may block connections from datacenter IPs or access other services that need to come from your IP.
Just go to Settings > Computer > Route egress to enable it. ( you may need to update Grok Bot's computer first )
Transparency and honesty in evals is critical, super proud of the work the team is doing with CursorBench even if it means our own model goes down a bit on the bench.
this is absolute genius. if you're reading this and you have a tech product, any product really, you should immediately put tinkabot on it and build an integration
meet my new grok @bot tinkabot (v0.1.0)! she helps you make high quality grok bot plugins. you can ask her to look at your APIs, create an MCP and/or skills for it, and then wrap that up as a plugin that you can then submit to us for approval.
once it's approved, your plugin will be available for all grok @bot users to use!
let me know if you run into any issues, i've used it on some small toy APIs but would love to see how well it works on real ones
x.ai/bot/br5f3C4mc75QCMEHasz…
Maxime Prades retweeted
we're sprinting to fix auth issues in grok @bot
please let me know if you run into any bugs getting access with your Cursor/SuperGrok subscription
Ok let’s try this again, calling all Bugbot power users. What are your biggest bugbot needs and feature requests?
This makes it all worth it 😂. Tell me we’re not living in wild times
I built a @cursor_ai client that runs inside World of Warcraft as an addon so I can keep an eye on my agents and prompt them when i need.
Maxime Prades retweeted
Cursor has been a trusted partner of Anthropic since Sonnet 3.5. We’ll continue to increase compute to support Claude models in Cursor and are excited for what comes next with them at SpaceX.
Maxime Prades retweeted
if you're a women in AI in SF, come join @spacexai & @a16z for our first ever grok bot build night!
learn about @poteto, @clairevo & @stuffyokodraws's workflows and set up your first AI teammates
join us! luma.com/a16zgrokbotbuildnig…
Maxime Prades retweeted
A friend of mine just moved to San Francisco and was completely lost in our vocab.
So I built this for her and everyone like her: sfisms.org
The field guide to SF vernacular.
Got an ism? Pitch it and an agent adds it after expert review by our panel :)
these are not some rookie numbers rohan! love me some poteto mode. nice work getting cloud agents rolled out effectively. love seeing those stories with @cursor_ai
our cloud agent usage at decagon is up 6x over the past 3 months. cloud agents are quickly becoming the default way our team works because:
- they scale beyond a single laptop: each agent gets its own isolated sandbox (no worktrees) so there’s no local resource contention
- you can spawn swarms from anywhere (web, slack, linear): designers and PMs are shipping
- they can run for hours, spin up all services, and put up PRs with screenshots/videos
but if verification isn't trustworthy, you just multiply slop. the bottleneck isn’t model intelligence, it’s the environment setup and verification. to verify effectively, you need to:
- give it the same tools an engineer has locally (same stack, MCPs, context)
- create a framework (like @poteto's pstack) that encodes how the agent should create/test its changes and ship artifacts you can review without reconstructing the session
the horizon of work you can give agents is much longer - cloud agents can pretty much ship while you sleep. probably the highest leverage thing an eng team can invest in right now.
coolest product i've worked on in a while! try it out, this is an early access program, so still lots of kinks to iron out before we're officially out
This team has been super helpful over the last few months, really feels like the "AI teammates" moment we've been talking about for years, the more specialized I made them (custom skills, personalities, memories and routines) the more helpful they got, Zidane who basically runs the team has refreshed context on what I work on multiple times a day and I find myself reaching for him more and more every day and every task.






















