Phil Cockfield
@pjc
π¦ββ¬<π΅> β πππ β /phil β github.com/philcockfield β
Ζ(π³) β 𧫠/sys β @cell <T>, π§ viva la small-tech (team human).
Design: "Human in the Loop" β
"4. Better planning process. Would be neat to see an agent that acts like a rigorous product / engineer thinker. I get that Claude Code can do this but I think exposing the creator to more decisions up front leads to a a much better initial product. Editing / curation matters!"
Unsolved vibe coding problems
1. Default aesthetics are too similar, e.g. Bootstrap (in a prior era) / Shadcn herding effect
2. Deployment flow with database, auth, payments
3. Domain names are still surprising hard! Subdomains are not consumer-grade. Vercel seems best here.
4. Better planning process. Would be ne
3
'speed to first illusion' vs. 'maintain'
(as complexity and entropy tear at you from all sides).
Greatness isnβt in the attain. Itβs in the maintain.
Anyone can create. Very few can sustain.
The biggest AI unlock for creators wonβt be making things, itβll be maintaining them.
Creation is being solved. Maintenance is the bottleneck for great creations.
I talk with young people all the time. Theyβre prompting and
3
Until agents can:
β’ Hold a stable, evolving theory of the system
β’ Track design intent across months
β’ Feel the cost of abstraction debt
β¦theyβll remain accelerators, not authors.
___
This will age like π's, but the hard part in there is the "taste" part. Does not seem impossible though on <?> time horizon.
1
the "corposlopβ’οΈ web" vs. the "sovereign web"
I agree with maybe 60% of this, but one bit that is particularly important to highlight is the explicit separation between what the poster calls "the open web" (really, the corposlop web), and "the sovereign web".
https://x.com/tom777kruise/status/2006710624424702362
This is a distinction I did not realize until rece
1
1