@beffjezos nah. any world which does not involve sovereign intelligence is one that must be avoided. they clearly want centralized control of this technology, there can be no reconciling.
biggest thing i have been thinking lately is questioning realtime as a requirement. google docs basically went through this massive pain of OT to get realtime editing, when in reality nobody really edits docs together like that. serious work is done more like a feature PR, you spend a bunch of time thinking by yourself and produce a coherent logical unit; think there is a generalizable principle there wrt knowledge work.
ofc would be untenable to have a massive repository (say an entire company's context checked out locally), but if you could specify a subset and the agent could just mediate the hydration and dehydration for you as you work. you could see having only a subset and even a subset of its history for working locally while having the desired properties on the remote
clearly a strong need for a new solution here.
seeing this my initial thought is that honestly a more scalable github with sparse materialization feels like it would be the best solution. editing realtime is really not a requirement for knowledge workflows, as changes tend to come in logical groups (transactions, i.e. commits).
Today we’re launching FlashDrive with @ycombinator 🚀
FlashDrive is a shared cloud drive where your team and all your AI agents work in the same files across computers and cloud sandboxes.
> Search the whole drive with grep, regex or code. Our custom indexing layer finds matches without downloading everything.
> Files stay in the cloud, and your agent only streams the part it needs.
> Start in Claude Code on your Mac. Pick up in Codex.
> Edit the same file at the same time. FlashDrive merges changes or keeps both versions.
> Connect over MCP or mount the drive on macOS or Linux. Give a sandbox access to just one folder for as little as five minutes.
Dalton and I are building the filesystem the next generation of companies will run on. And we're calling it @flashdrive_ai
↓ Try it free! Link below
market’s gonna be in for a massive shock when they realize that a deployable self learner that doesn’t need to train on the entire internet could evaporate the labs’ moat overnight..
one cannot help but notice the irony of the names of the frontier labs, we should colloquially start calling them their real names:
Anthropic → Misanthropic
OpenAI → ClosedAI
let us have a little honesty.
furthermore, i consider that Misanthropic must be stopped.