Today, we're launching shift. We're starting by cleaning your apartment in New York City, for free.
Here's how it works. Book a shift cleaning. A vetted shift operator comes to your home wearing one of our devices. They clean. They leave. You pay nothing.
In exchange, we record the cleaning. Robotics is being built on data about how people do daily tasks, and the value of that recording is what funds the service. Anything personal in it is anonymized before the recording is processed.
By now, you have heard about the shift to AI more times than you can count. About the shift toward you, the part where you actually feel it, you have heard almost nothing. Shift is what starts to make it concrete, in specific cities, with specific services.
Today, cleaning in New York. Soon, handymen, repairs, and errands across the globe. And this is just one side of shift, with more on the way.
Comment “shift” and we’ll send you an early access link.
Also worth noting out that the 'useful consumer features' he's waiting for are rarely the ones that get left on the cutting room floor
More often than not it's refactoring, security, performance, etc. projects
If ai is helping Uber work through that backlog then it could still be a great return on spend
(Especially given all the exciting new critical vulnerabilities that keep popping up. Thanks, Mythos.)
Maybe my boomer is showing but this makes about as much sense as people texting each other from the same room
Transcribing for notes is one thing but if you're going to intermediate each other with AIs then why have the meeting at all
You can now transcribe meetings in real time using Codex and ask Codex questions about meetings as they're happening!
I updated my new Codex Meeting Recorder skill to use GPT Realtime Whisper. Tell Codex to use the skill, and it will start transcription and show it in the preview pane. You can ask questions about the transcript as it's being generated. When the meeting's done, tell Codex to end it, and you'll get the full transcript and a formatted version (the formatted version needs work; on my list).
Note that because this uses the GPT Realtime Whisper endpoint, it's more expensive than transcribing at the end of the meeting, at $0.017 per minute (so about $0.51 per 30-minute meeting). I'm considering adding a local realtime option in future, such as with Nemotron Speech Streaming.
GitHub link is in the thread along with some instructions.
@maggielove_ I'm hardly an expert here but to paraphrase my wife (who much closer to one) the issue is very few internal review boards will greenlight new research on women's reproductive health
Sidestepping this w/ self-experimentation & sharing a la Bryan may actually be the workaround
The goldilocks zone of AI use is being far enough away from the work that you're good at delegating it
But close enough to the work that you can sense when the AI is bullshitting you
@timyoung It’s the bless your heart of enterprise software
10% of the time they’ll actually put it on the roadmap but the other 90 it means that idea was dumb but I’m too polite to say it
Is this not an open secret?
We’re now entering the ROT* era of software development
Where as foundational model subsidies fade, leaders start to ask what quantifiable results their unlimited token spend is buying them
Nailing down these benefits (and culling negative roi spend) is the next hurdle to clear
I can only think of a handful of companies poised to do it
(*return on token)
new favorite AI workflow (scheduled to run daily):
9:00am
- pull yesterday's granola notes into local .md files
9:15am
- fetch my recent activity across slack, email, docs, linear, etc.
9:30am
- find all commitments that I made to others
- find all commitments that others made to me
- send me a slack msg summary of the new and resolved items
We've been trying to track down a high CPU usage issue on Sero
Turns out @Meta 's AI bot was flooding our endpoints with every possible permutation of query params, slamming our server despite robots.txt policies against
Anyway, here's what happens when you block them fully
Solid read + backed up with what I see in Argentina
Everything crypto is USDT on Tron, basically no other chains in use for non-speculative use case
But it's used almost entirely for cross-border transfers and a savings account
All intra-country transfers are made on traditional financial rails via 'transferencia' (effectively venmo-style instant transfers to public aliases)
Given how well this serves 99.9% of use cases I struggle to see Argentina moving culturally to crypto transfers for daily purchases
With that said, the savings account angle is non-rivial
Argentina is the most multi-currency country I've ever spent time in with average people very accustomed to a three tier (real estate > dollars > pesos) approach to wealth storage
Entirely possible that stablecoins catch hold and largely replace the stacks of physical dollar bills floating around the country for medium-sized savings amounts — especially given Argentina's history with capital controls and people's general distrust of the govt for anything financial
Average lending rates have dropped to 2.9% - lower than the 3-month U.S Treasury rate.
Onchain credit is filling the gap, providing $300Bn+ in stablecoins with yield-generating opportunities.
@solofunk dove into the data in our first research report on Onchain Credit.