PWV Portfolio Feed

:( What would you have expected too see?

git forges like github or gitlab challenged the ci space by building solutions where you can bring first and third party runners

the result? cheaper and better ci through companies that provide you with runners

and best of all? llm tools like “claude code” can rewrite the pipelines for you

the watercoloring-to-pagerduty-alert sunday afternoon vibe shift

safrà21 has easily become by favorite lunch place before taking a flight from mallorca

i barely write code these days 😅

i thought llms would stop our innate obsession for taking the language that we love to other platforms and instead choose the language of the platform, but i was wrong

swiftui previews would be an amazing feedback tool for agentic coding tools

however, their unreliability, slowness, and proprietary nature makes them as terrible for llms as they are for developers

apple is paying for all their years investing in proprietary solutions

i’m surprised no one has agentic-coded a buffer alternative with native ios client, and maybe local-first?

the diff rendering could be improved, i'm thinking about making it a word-wise diff, any other ideas for diff algorithms lmk, i want to know more about diffs

good update

leaning on proprietary undocumented formas like xcode projects might have seemed a good idea for many years, but it’s become as painful for llms as it’s always been for humans

ai has made apple release cycles feel terribly slow

by the time they release something to catch up with the industry, the idea is already obsolete

apple has traditionally made app development transitively dependent on macos and apple hardware through xcode, but in a world of agents, or headless coding, there’s never been a better opportunity to break that dependency further

a macos environment for the agent can be an implementation detail

with ai writing automation for you in bash, and companies providing runners for your ci jobs, may abstractions and concepts such as mobile ci fall apart

damn! even xcode projects

ai managing them means we can gear tuist’s mental energy towards shaping new app building experiences

if you’ve got an agent building features remotely (e.g., for an ios app), how important it is to go through the standard pr review process?

won’t we get to a point where we can trust the agent to review the work and also run automated tests such that work can land directly on “main”?