@clairevo Yea lots of cool solutions popping up for for this. We've been solving for "mini-apps" and supporting multiuser editing/collab and permissions building https://t.co/uFfhrzhSw9.
Just something about putting mini-apps and presentations on a full netlify setup feels heavy/complex.
@paulg IRL or online, I think with how much info gets sold, breached, etc the strategy becomes persistent disinformation. I’m answering wrong for almost anything if I can
That Stripe PM always does the same on here- are we saying Stripe has no clue how to make Atlas better? I get that it feels lazy to just ask “what do you want”. I truly get the reaction of wanting someone to have vision and an opinion rather than crowdsource a roadmap. But if you’re not actively looking for every piece of feedback possible… you’re missing inputs that will sharpen your vision and yes even worthwhile features.
Poor Boris at Anthropic is trying his hardest, but the fact he has to ask how they can do better goes to show Anthropic has literally no clue why people are ditching Claude Code for Codex.
@mher_ai@trq212 We were doing this and wanted more flexible sharing/collab. Built this little tool so we can push htmls from Claude code to live shareable links via MCP https://t.co/l94yVbfakE
@Top100Rick Was taught this in a lesson and it feels odd but is surprisingly easy to pickup. The one downside is if wet, will dig in to the turf rather than bounce.
I’m sure there’s a wet ground grind I could be switching to. Just surprised me.
Mostly agree. Seeing a lot of Eng move from Claude code to Codex tho. Claude “being nerfed” a semi fumble that let OAI back in.
Switching costs seem to be near 0 which is wild at this level of investment.
To your point, I think Claude code has the brand in coding and folks will move back and even stay assuming the nerf thing is solved and models are even in capability.
So could be that OAI got drawn into the coding battle with some false hope. Could also play out differently…
@gerstenzang@sbensu no revenue? show users
no users? show retention
no retention? show evals
no evals? show latency
no latency win? show LOC
no LOC increase? show tokens burned
no tokens increased? show prompt length
“Children from sibships of 1–3 typically left less wealth than they inherited. But children of families of 7–12 left nearly twice as much wealth as they inherited.”
Fascinating since larger sibship size usually predicts lower IQ, lower education, lower income.
So is it that large sibship in wealthy families inherit the platform without the cushion? They have every advantage except $$$$. high social capital, low financial capital.
Raise your kids hungry but give them every tool, I suppose.
The wealthiest wealthy families substantially converge to the mean across generations, but nevertheless remain very wealthy relative to merely rich families and the population at large.
They can be expected to remain above-average for ~10-12 generations.
one of our most fun levers to drive claude code adoption internally has been a leaderboard for token burn. It leads to two really effective behaviors
1. Competition: Heard on the daily now "I can't let so and so beat me today"
2. Shame: Anyone low on tokens for the day is getting screenshotted in slack by their haters
the technology that kills a job never does the job better.
ATMs did exactly what tellers did at 27 cents a transaction vs $1.07. teller employment went up. then the iPhone came along, had nothing to do with banking, and teller jobs dropped from 332k to 164k in twelve years. (h/t @david_oks) https://t.co/Xepx3bIh1Z
the ATM fit into the branch. the iPhone deleted the branch.
many companies trend toward making ATMs. Its easier to replicate a workflow that exists and has a market. But talking about how you can delete workflows and what that could enable is much more interesting...
Claude code et al is maybe deleting the product team? Giving rise to the builder? We're in the middle of the shift now and its remarkable how debatable the outcome is. It isn't clear which "ATMs" will die and how quickly.