@vladsiu It means two iPhones can share the same grocery history basically built for couples / two-person households.
Thanks for catching that. “Made for two” is definitely too vague I’ll rework it 😅
@doodlestein@_junhoyeo My 600B tokens suddenly look pretty modest next to this lol.
Huge respect though this is an absolutely insane amount of building. Love seeing people push agentic coding this hard.
I still can’t get used to ChatGPT and Codex being in the same app.
Am I the only one who thinks it was actually more convenient when they were separate? lol
@alexcloudstar I use both heavily. Codex is my default now much more generous limits, and SOL is noticeably better at long, complex tasks. It stays on track and gets lost in context far less often.
Claude still wins at design though.
Codex limits are way more generous, and SOL is much better at long, complex tasks. It stays on goal and gets lost in long contexts far less often.
The one area where Claude is still objectively stronger is design - Design Claude is insanely good.
But SOL is closing that gap fast. Back in the GPT-5.4/5.5 days, the difference was almost comical.
Wouldn’t be surprised if Astra makes Codex better at that too.
@danellisona My account is 16 years old, but I barely posted for the last 12 years.
Somehow it feels like growing an old inactive account is even harder than starting a new one lol
Getting Radish ready for the UK 🇬🇧
We taught it pounds, UK dates, and collected 94 receipts from 10 stores for testing.
We also fixed our stats so we can see where users get stuck.
Building an app is 50% code and 50% explaining that 02/09 is not February.
Thinking about building this:
Submit your SaaS → get 10–20 real builders to try it and give you feedback.
You can earn tests by testing other people’s products, or just pay if you don’t have the time.
Before I write any code, I want to validate the idea:
Would you submit your project to this?