parasocial relationships are a hell of a drug thats not to take away from the mans achievements but i do sometimes think the admiration that some people get publicly can actually paradoxically shade their actual prowess what i think i mean is that expectation and reality are in my experience anyway, usually at least somewhat mismatched... but that goes both ways (the good way and the bad) its my whole problem with the social 'following' paradigm, especially where it overlays with real skill, things that require actual mastery its so easy to conflate credibility with the consensus of credibility, clearly in this case it is deserved but its dangerously easy to get wrong and i think what my point is is that, conceptually, the fact it exists leaves it open to interpretation on the flip side im a big believer in giving people their flowers, i love celebrating people, its such an important part of life, friends, family, strangers... maybe its the scale that makes it weird, maybe its the perception mismatch or the information decay along the way... i personally had a strange, unsettling feeling when I saw Sunil was leaving Cloudflare and i have 0 personal connection to him. maybe its our innate disdain for change and uncertainty, maybe theere's something to it? maybe i just need to go to bed...

how we interact with software is changing, when i first build FounderMail it didn't have a UI... with agents taking on more and more of our daily tasks its tempting to remove our input all together but for now im settling on building products agent first. i see agent first as an evolution to the mobile first revolution we had back when i still had some sort of concept of time, everything is moving so fast now. give your agents this url and let me know what they think!

I mean you don’t NEED anything… the thing people don’t seem to get about Cloudflare’s developer platform is the value comes from all of the stuff you DONT NEED to do when building on it.

'be more ambitious' i regularly get laughed at by a fellow engineer friend of mine for saying; 'i could literally build anything, now i want to build something that really matters' but i genuinely (within reason) believe that to be true. i like this talk not because of what @mattzcarey @thomas_ankcorn and @dillon_mulroy built (ive used it since closed beta, its good) but HOW it was built. its not just vibe coding, its modern, agentic software engineering, that's so exciting. be more ambitious isnt just blah blah conference speak, i really believe it to be something every engineer should be embracing. 0 surprise to me that 1 of the 2 questions there was time for was about 'writing code is dead'. Over the last 15+ years ive watched how intimately intertwined writing software and being good at solving problems can be with an engineers ego (not a dirty word), sense of self and purpose. its obvious that the perceived ripping away of one of the core parts of the role would be a tricky transition for a lot of people. i think we have to be sensitive of that and try our best to encourage the industry to embrace the change with cautious optimism, for the collective good of the industry. it really is such an interesting time to be alive.

coming from doing C# and PHP where i dont htink monorepos make as much sense, it took some convincing when i picked up TypeScript a couple of yeas ago, to actually see the value (to the extent of doing multi and migrating it to mono, twice). now ive speed run multiple HUGE TS projects in monos, i still see the value BUT last week i took a stab at Cloudflare's multi agent github replacement project, knocked out 1 million lines of code in the 4 days i had to think about and build the thing. i dogfooded my solution on my solution, as soon as it was possible. i realised while i had actually solved parts of the problem (the solution is legitimately nice to use and faster than big merge queues), id not gone deep enough (obviously, you only spent 4 days you idiot). ive spent the the last week with some threads i still want to tug on. about an hour ago i was thinking 'i think monorepos might actually be the problem, but im not sure just going multi is the answer either' we have to be prepared to shatter any and all existing paradigms at every level of the software lifecycle if we are going to get 'this' right, that doesnt mean re inventing the wheel but at least being prepared to make it out of different stuff. im going to tug on those threads as soon as possible. couple of projects to a launch then im on it

Open AI or Anthropic, roll up, roll up, place ya bets

some times i read posts like this and remember not everyone is building on Cloudflare, crazy. (no they dont pay me)

Gee, Brain, what are we gonna do tonight? The same thing we do every night, Pinky, try to take over the world!

I need to get rich because I want loads of donkeys and horses, not like in a stable to ride like wild fucking horses, majestic beasts (save this post for when I build something sick then exit)

I make Claude use PhpMyAdmin in the browser so he knows what made me who I am today

BRR just going to patch my PHP servers via FTP

while i had fun putting this together, i did find myself thinking 'does any of this even matter?' what would an agent want? is sensible, human readable code even important now? im someone that has always valued semantically meaningful code and if im honest its been a while since ive really invested in it, invested in the way i wold have with every single line of code i ever wrote. for now, im going to lean towards yes, im going to release this as a simple, library that allows you to meaningfully combine and compose system one and LL models. that isnt going to stop me from taking it further and thinking even more about how i can make this attractive for agents, and if this isnt, what is.

My Durable Edgeling and his spawn both died... I guess they weren't so durable after all