I built FlowIn:
→ autocomplete in every app on your Mac
→ one-click replies, rewrites, and translation
English is my second language, and the gap between what I mean and what I type costs me hours every week.
Most writing tools fix what you already wrote. FlowIn works one step earlier.
Free to download at https://t.co/qKPNaZh1lJ. Would love for you to try it!
Results and findings of this paper:
Physical activity and video gaming have completely different benefits. Spending time playing video games was associated with significantly better performance in thinking skills, but it was not related to mental health scores. https://t.co/Z2F8UN3sOM
I'm using this systen-wide autocomplete to save 50% of the typing every day.
It feels like mind reading
Things like this should be a default on your computer.
Hey yall, quick announcement:
Applications are due soon for Character Labs, our startup accelerator in San Francisco! More info and application link at https://t.co/vvU4hN0s1N
The only accelerator that's 100% focused on product-market fit, it's based on sprints that we've run with 300+ teams over the past 20+ years (including those at Google, YouTube, Flatiron Health, Slack, Uber, One Medical, Gusto, Phaidra, Inductive Bio, Fathom, Orbital, House Rx, and Extend).
If you're pre-pmf and interested in working with us, please consider Character Labs — and reply/dm if you have any questions.
I did office hours today with a startup that had three different paths to get huge. I told them to optimize for growth rate and let that decide. Maybe the winning path will be a fourth one we don't know about yet.
I love agentic coding harnesses, but they shouldn't be primarily terminal-based. The terminal is great for quick and precise commands, but information density is extremely low and UI affordances are minimal. Maybe provision of TUIs is worthwhile for occasional use (when establishing a tunnel is too annoying, or something), but it feels very strange for this to be the default modality. It took a long time for dynamic language REPLs to break out of the terminal (Jupyter notebooks and similar); I hope we don't have to wait as long for the harnesses.
Last month, @Azure acknowledged a billing issue that overcharged us by at least $60K for GPT-5.6 usage. We were not alone: many customers using Azure-hosted GPT-5.6 models were significantly overcharged between July 9, 18:00 UTC and July 22, 18:44 UTC.
A full month later, the incorrect charges still appear on our latest bill. We have followed up every day, yet its billing team still has not corrected the wrong overcharges.
If this happened at Alibaba or Tencent, leadership would be held accountable, possibly removed. At Azure, nobody seems to care.
I’m beginning to understand why large U.S. companies often struggle to compete with startups, and why Chinese tech giants may eventually surpass their U.S. peers.