A widely discussed idea on X: on Oct 31 Stripe CEO Patrick Collison (@patrickc) posted that he is building an "improvements" feature into all of his personal vibecoded applications, a freeform text input hooked up to an agent that implements whatever is submitted. He says another person who uses one of the apps has already implemented 150 improvements in a few days, and he floats the name "liquid software" for the pattern. He is also candid that he does not know whether it scales: whether it becomes unmaintainable, how collaborative software works this way, and whether it can be secure and reliable. The post drew about 2,130 likes, 1,900 bookmarks and 147K views. This is a write-up of a personal opinion and the discussion around it, not an official Stripe product announcement.
Key Takeaways
- ✓Buzz: about 2,133 likes, 1,896 bookmarks, 217 replies, 76 quotes and 147K views (public X metrics as of Nov 1, 12:15 SGT)
- ✓Pattern: every personal app gets a freeform text box wired to an agent that yolo implements whatever is submitted, which makes sanding a small rough edge almost free
- ✓Data point: Collison says another person using one of his apps implemented 150 improvements in a few days (self-reported; the app itself is not public)
- ✓His own open questions: does it become unmaintainable, can model gains outpace feature appetite, how does collaborative software work, and can it be secure and reliable
- ✓Debate: supporters say it will become standard and could even ship bespoke per-user features behind flags; skeptics note conflicts start once a second user asks for the opposite change, and maintenance and security remain unsolved; Collison himself compares it to the user-modifiable visions of Lisp machines, Smalltalk and HyperCard
Key Decision Metrics at a Glance
Turn your technical choice into a development budget
Compare 40 dev plans & simulate token costs vs $20/mo subscriptions
Project Links & Resources
Direct AccessIn-Depth Technical Analysis
Background and the problem
Coding agents have made it cheap for one person to build small apps, but once an app exists, fixing a small annoyance still means opening an editor, describing the change and waiting for an agent. On Oct 31 Stripe CEO Patrick Collison (@patrickc on X) shared in a long post how he handles this in his own personal apps, and it quickly became one of the most discussed AI posts of the day. To be clear, this is his personal experiment and opinion, not an official Stripe product announcement.
Architecture and how it works
As the post describes it, he is adding an "improvements" feature to all of his personal vibecoded applications: a freeform text input hooked up to an agent that yolo implements whatever is submitted, with no planning or review step. He says this makes the activation energy for sanding a rough edge or adding a feature extremely low, and that another person who uses one of his apps has already implemented 150 improvements through it in a few days. He floats the name "liquid software" for the pattern and says apps without it now feel slightly rigid and anachronistic.
Benchmarks and measured numbers
This is not a model or product launch, so there are no benchmarks. The only verifiable numbers are the post's public metrics: as of Nov 1, 12:15 SGT, about 2,133 likes, 1,896 bookmarks, 217 replies, 76 quotes and 147K views. Bookmarks nearly matching likes suggests many developers saved it as an idea to come back to. The 150 improvements figure is self-reported, and the app in question is not public.
Getting started
Collison raises open questions himself: whether the approach becomes unmaintainable over time (perhaps model improvements outpace his feature appetite), how collaborative software works in this model, and whether software designed this way can be secure and reliable. His conclusion is that it works very well in a personal context, and he compares it to the user-modifiable visions of Lisp machines, Smalltalk and HyperCard. Quote posts split roughly two ways: some call it the future of software, even imagining bespoke features generated per user and shipped behind feature flags; others point out it is ideal with one user, but conflicts begin once a second person asks for the opposite change, and maintenance and security still need extra guardrails. If you want to try it on a personal project, the key is giving the agent solid tests and a way to roll back, and limiting it to apps where you can accept the risk.
Benchmark side-by-side against alternatives, or calculate monthly token cost vs subscription break-even.
Discussion & Comments
0Sign in to join the discussion
Connect with AI developers to exchange benchmark insights.