one of the secrets to the design improvements at @cloudflare is that it wasn’t done by committee
1. hire (or acquire) folks like @burcs@nandafyi@mattrothenberg@BraydenWilmoth
2. give them big challenges
3. get out of their way
4. support them by unblocking where needed
re: cloudflare emerging as the agent cloud of the coming century
our "r&d" department (aka ETI, Workers Org) is weird in the truest, most wonderful sense of the word. we ship (early, often) and support customers in prod, we build in public (and don't stfu about it), we're unafraid of making weird bets, we learn and iterate a ton, work directly with users/customers (in tons of slack channels, DMs, mails, all that) operating at a scale that should make this all impossible. there are _obvious_ tradeoffs, but I don't see any other org/company willing to even try, let alone commit to those tradeoffs
I think it's based on matthew/michelle being clay christensen's students in HBS (and them continuing the relationship for years later), kenton caring about the technical definition of disruption theory, and dane being the most high signal individual I've ever known; probably has a lot to do with it
@ritakozlov@CherryJimbo It would be nice to see D1 meet the needs of a site like the Cloudflare blog? I imagine there are plenty of other folks with similarly large sets of data they'd love to put in D1 instead of somewhere else?